Professional Documents
Culture Documents
Trends
2021
Trending
Trending thetrends:
the trends:Twelve
Twelve years
years ofofresearch
research
2
CORE TRUST BUSINESS OF TECHNOLOGY CLOUD & DATA AND ANALYTICS & DIGITAL EXPERIENCE
MODERNIZATION DISTRIBUTED PLATFORMS ARTIFICIAL INTELLIGENCE & DIGITAL REALITY
2020
Ethical Horizon Global study: Finance & Architecture Digital Human
technology next The kinetic the future awakens twins experience
& trust leader of IT platforms
2019 DevSecOps Beyond Connectivity NoOps in a AI-fueled Intelligent Beyond
& the cyber the digital of tomorrow serverless organizations interfaces marketing
imperative frontier world
2018 The new Risk Exponentials CIO survey: No-collar Reengineering API Blockchain to Enterprise Digital
core implications watch list Manifesting workforce technology imperative blockchains data reality
legacy sovereignty
2015 Core Risk CIO as chief IT worker Software- API Amplified Ambient Dimensional
renaissance implications Exponentials integration of the defined economy intelligence computing marketing
officer future everything
2013 Reinventing No such Design as CIO as Business IPv6 (and Finding the Mobile only— Social Gamification
the ERP thing as a discipline postdigital of IT this time face of and beyond reengineering goes to
engine hacker-proof catalyst we mean it) your data work
2012 Outside-in Digital Measured Hyper-hybrid Big data Geospatial Enterprise User Social
architecture identities innovation clouds goes to visualization mobility empowerment business Gamification
work
TABLE OF
CONTENTS 4 13 64 116
Letter from Strategy, MLOps: Rebooting the
the editors engineered Industrialized AI digital workplace
6 29 82 135
Executive Core Machine data revolution: Bespoke for billions:
summary revival Feeding the machine Digital meets physical
10 47 101 150
Macro technology Supply Zero trust: Never DEI tech:
forces unchained trust, always verify Tools for equity
162 Acknowledgments
1 2 3 4 5 6 7 8 9 Letter from the editors 4
A
s poet Robert Burns mused, the the face of change. We have seen countless, swimming naked when the tide went out.
best-laid plans of mice and men often inspiring examples of resilience this past year Likewise, executive-level planning discussions
go awry. In January 2020, most of us as organizations and entire sectors assessed about the future of work had been just that:
had plans—thoughtful road maps to guide their circumstances, revised their strategic about the future. The pandemic crashed
our organizations, our technology, and our plans, and marched toward the future. comfortable schedules from years into weeks.
lives through the months to follow. And then
COVID-19 punched the entire world in the We anticipate that for most, the future they find With that background, this year’s Tech Trends
mouth, rendering useless many of these best- will differ markedly from the realities of January report discusses the opportunities, strategies,
laid plans. Seemingly overnight, a strange, 2020. The COVID-19 crisis has driven change in and technologies that will drive new plans
historic event disrupted our assumptions and an important and unexpected way. A growing during the next 18 to 24 months and beyond:
forced us, with a shocking degree of urgency, to number of organizations across sectors
become more adaptable and responsive than are accelerating their digital transformation • For enterprise technology, we spotlight
we had thought possible. efforts not only to make their operations the importance of aligning corporate
nimbler and more efficient but to respond to and technology strategy; we revisit the
Mindful that the pandemic’s impact continues dramatic fluctuations in demand and customer critical core and how digital nonnatives
to ripple across societies, markets, and lives, expectation. For example, while many supply are using cloud, low-code, and platform-
we present Tech Trends 2021. The theme of this chain leaders were confident of their ability first strategies to juice legacy assets; and
year’s report is resilience. To us, this means a to function during disruptions, we found out, we take a deep dive into supply chain
stubborn determination to adapt and thrive in as Warren Buffett once quipped, who was transformation.
1 2 3 4 5 6 7 8 9 Letter from the editors 5
Executive summary
Case studies, insights, and the trends
Strategy, engineered MLOps: Industrialized AI Rebooting the digital workplace
• Joseph Fuller, Harvard Business School • National Oceanic and Atmospheric Administration • Thomson Reuters
• Peter Schwartz, Salesforce • Morgan Stanley • Lloyd’s of London
• Anthem • JLL
Core revival • Swami Sivasubramanian, AWS • Dan Torunian, PayPal
• Albemarle
• Sogrape Machine data revolution: Bespoke for billions:
• GM Financial Feeding the machine Digital meets physical
• Justin Kershaw, Cargill • AT&T • Hans Neubert, Gensler
• Loblaw
Supply unchained • ABN AMRO DEI tech: Tools for equity
• Pactiv Evergreen • Lutz Beck, Daimler Trucks North America • Deloitte US
• John Tomblin, PhD, Wichita State University
Zero trust: Never trust, always verify
• Takeda
• Halliburton
• John Kindervag, Palo Alto Networks
1 2 3 4 5 6 7 8 9 Executive summary 7
Technology today is a source Modernizing legacy enterprise Long considered a cost of doing
of new competitive advantage systems and migrating them business, supply chains are moving
for some organizations and a to the cloud may help unleash out of the back office and onto
threat to ongoing survival for an organization’s digital the value-enabling front lines
others. As a result, the distinction potential. Until recently, these of customer segmentation and
between corporate strategy undertakings could also bust product differentiation. Future-
and technology strategy is blurring—each needs to that same organization’s digital transformation budget. focused manufacturers, retailers, distributors, and others
inform the other. Savvy corporate strategists are looking For many, the cost of needed cloud migrations and other are exploring ways to transform the supply chain cost
beyond their organization’s current tech capabilities and core modernization strategies can be prohibitive. This center into a customer-focused driver of value. They are
competitive landscape to consider a broader range of is about to change. In what we recognize as a growing extracting more value from the data they collect, analyze,
future possibilities about how technology can expand trend, some pioneering companies are beginning to use and share across their supply networks. Finally, some of
where they play and how they win. But the complex range clever outsourcing arrangements to reengineer traditional these organizations are exploring opportunities to use
of uncertainties and possibilities can be too much for the business cases for core modernization. Likewise, some are robots, drones, and advanced image recognition to make
human brain to process on its own. That’s why strategists exploring opportunities to shift core assets to increasingly physical supply chain interactions more efficient, effective,
are turning to strategic technology platforms equipped powerful platforms, including low-code options. Finally, and safe for employees. Granted, transforming established
with advanced analytics, automation, and AI. Organizations many are advancing their platform-first strategies by supply chains into resilient, customer-focused supply
are using these tools to continually identify internal and addressing technical debt in ERP systems and migrating networks will be a challenge, and for most organizations,
external strategic forces, inform strategic decisions, and nonessential capabilities to other platforms. In a business it will be an ongoing journey—one of critical importance.
monitor outcomes. As a result, companies are transforming climate defined by historic uncertainty, these innovative The kind of disruption we have seen with the COVID-19
strategy development from an infrequent, time-consuming approaches for extracting more value from legacy core pandemic could well become the norm. When the next
process to one that’s continuous and dynamic, helping assets may soon become standard components of every global event hits, technology and supply chain leaders won’t
strategists think more expansively and creatively about the CIO’s digital transformation playbook. be able to claim they didn’t see it coming.
wide range of future possibilities.
1 2 3 4 5 6 7 8 9 Executive summary 8
MLOps: Industrialized AI Machine data revolution: Feeding the machine Zero trust: Never trust, always verify
Sophisticated machine learning With machine learning poised to Sophisticated cyberattacks and
models help companies efficiently overhaul enterprise operations shifting enterprise environments
discover patterns, reveal and decision-making, a growing have undermined the traditional—
anomalies, make predictions number of AI pioneers are and somewhat flawed—castle-and-
and decisions, and generate realizing that legacy data models moat approach to cybersecurity.
insights—and are increasingly and infrastructure—all designed Zero trust is rooted in the concept
becoming key drivers of organizational performance. to support decision-making by humans, not machines— that modern enterprise environments necessitate a different
Enterprises are realizing the need to shift from personal could be a roadblock to ML success. In response, these approach to security: There’s no longer a defined perimeter
heroics to engineered performance to efficiently move organizations are taking steps to disrupt the data inside which every user, workload, device, and network
ML models from development through to production and management value chain from end to end. As part of is inherently trusted. In zero trust architectures, every
management. However, many are hamstrung in their efforts a growing trend, they are deploying new technologies access request should be validated based on all available
by clunky, brittle development and deployment processes and approaches including advanced data capture and data points, including user identity, device, location, and
that stifle experimentation and hinder collaboration among structuring capabilities, analytics to identify connections other variables that provide context to each connection
product teams, operational staff, and data scientists. among random data, and next-generation cloud-based and allow more nuanced, risk-based decisions. Data,
As AI and ML mature, a strong dose of engineering and data stores to support complex modeling. Together, these applications, workloads, and other resources are treated
operational discipline can help organizations overcome tools and techniques can help organizations turn growing as individual, manageable units to contain breaches, and
these obstacles and efficiently scale AI to enable business volumes of data into a future-ready foundation for a access is provided based on the principle of least privilege.
transformation. To realize the broader, transformative new era in which machines will not only augment human The automation and engineering required to properly
benefits of AI and ML, the era of artisanal AI must give way decision-making but make real-time and at-scale decisions implement zero trust security architectures can help
to one of automated, industrialized insights. Enter MLOps, that humans cannot. strengthen security posture, simplify security management,
also known as ML CI/CD, ModelOps, and ML DevOps: the improve end-user experience, and enable modern
application of DevOps tools and approaches to model enterprise environments. But the move to zero trust could
development and delivery to industrialize and scale require significant effort and planning, including addressing
machine learning, from development and deployment to foundational cybersecurity issues, automating manual
ongoing model maintenance and management. processes, and planning for transformational changes to the
security organization, the technology landscape, and the
enterprise itself.
1 2 3 4 5 6 7 8 9 Executive summary 9
Rebooting the digital workplace Bespoke for billions: Digital meets physical DEI tech: Tools for equity
As the world’s largest unplanned When we look back, 2020 will Many organizations are embracing
work-from-home experiment likely be the turning point when diversity, equity, and inclusion
continues, many business leaders most of the population adapted to as business imperatives, with a
are asking as yet unanswered digital interactions to conduct their growing number adopting holistic,
questions: When the dust settles, everyday lives, whether working organizationwide workforce
will remote work become the from home, online schooling, or strategies that address biases
rule or the exception? Is a permanently remote workforce ordering groceries. Yet the prevalence of digital interactions and inequities to enhance enterprise and employee
sustainable? How will productivity and employee well-being has left many of us pining for the days of in-person performance. While HR professionals often lead DEI
be affected? Will innovation suffer in the absence of face-to- interactions. As we look to the future, we expect consumers strategies, technology leaders play a critical role as a
face peer connections? What will be the role of the physical will no longer be satisfied with distinct physical or digital strategic partner by designing, developing, and executing
office? Companies may be able to overcome the digital brand experiences: They will expect a blend of the best of tech-enabled solutions to address increasingly complex DEI
workplace’s deficits and ambiguities by more intentionally both—highly personalized, in-person experiences without workforce challenges. Over the coming months, we expect
embracing its positive aspects, including the data sacrificing the convenience of online transactions. In the enterprises to adopt new tools that incorporate advanced
generated by workers’ tools and platforms. This can help next 18 to 24 months, we expect in-person and digital analytics, automation, and AI, including natural language
organizations optimize individual and team performance and experiences to become more seamless and intertwined. processing and machine learning, to help inform, deliver,
customize the employee experience through personalized Online and offline interactions will not be separate and measure the impact of DEI.
recommendations, enabling remote work to be far more experiences anymore—the customer’s journey will be made
than a diminished proxy for the traditional office. And as up of in-person and digital elements that are integrated
onsite workspaces and headquarters evolve, organizations and intentionally designed to create a seamless brand
can use this data to create thriving, productive, and cost- experience that’s tailored to fit the individual customer’s
effective offices that are seamlessly interwoven with the behaviors, attitudes, and preferences.
remote experience.
1 2 3 4 5 6 7 8 9 Macro technology forces 10
F
or more than a decade, Tech Trends has and to the technologies beyond the horizon, These enabling technologies have already
explored the landscape of emerging the macro forces framework shows the powered a decade of disruptive change—
technologies and sought to understand inextricable links between the defining and they have a lot more innovation
their impact on business strategy. Within this technologies of the past, present, and future. mileage left in them.
broad landscape, we have identified several
These powerhouse
macro technology forces as the backbone • Disruptors. Digital reality, artificial
of business innovation and transformation. intelligence, and distributed platforms are
Discussed last year in the chapter Macro
technology forces, these powerhouse trends
trends distill 12 years of here today and are on the path to broad
adoption, sparking rapid disruption along
distill 12 years of research into an easily
consumable framework and taxonomy that can
research into an easily the way. Disruptors are currently shaping
the next decade of business innovation.
help organizations make sense of emerging consumable framework
technologies in order to articulate business • Horizon next. We expect horizon
direction, make strategic decisions, and and taxonomy. next technologies to mature over the
prioritize technology investments. coming decade and shape business
• Enablers. Digital experience, data and and technology strategies of the future.
Progressing from enabling technologies of analytics, and cloud have spawned multiple Ambient experience envisions ubiquitous
the past decade to disruptors of the 2020s innovative business models and strategies. interfaces, seamlessly integrated into the
1 2 3 4 5 6 7 8 9 Macro technology forces 11
environment, that anticipate and meet that combine insights and understanding
human needs. Exponential intelligence to recognize not only correlation
builds on artificial intelligence, adding but causation.
the capacity to recognize and respond
to human emotions, understand the • Computation. As computation abilities
external environment, and perform any scale, the long-term goal is abundance—
task. Meanwhile, quantum harnesses the the limitless ability to work with and gain
properties of subatomic particles to benefits from technology and information.
solve problems that are too complex
for today’s supercomputers. The macro Finally, to scale sustainably, emerging
forces framework also helps visualize the technologies must rest on a solid
evolutionary path of these technologies as foundation comprising: the business of
they influence interaction, information, and technology, the evolution of the IT function;
computation. trust, encompassing risks related to cyber,
regulatory, and ethics; and core modernization,
• Interaction. As human-technology the organization’s approach to rejuvenating
engagement continues to increase, the legacy core systems.
end game is simplicity—transparent and
natural interactions.
FOUNDATION
Strategy, engineered
FUTURE-READY STRATEGY
As strategy and technology become increasingly intertwined, Use strategy support platforms to identify driving forces,
strategic intent must inform technology choices. inform choices, and monitor execution and outcomes.
1 2 3 4 5 6 7 8 9 Strategy, engineered 14
TREND 1
Strategy, engineered
Tech-enabled business strategy, optimized for agility
And when asked about their strategic priorities strategy development. Nearly half (45%) refresh critical to effective strategy development and
and ability to execute, CSOs with disruptive, their strategy annually, or even less frequently: execution. In collaboration with the CIO, the
digital ambitions lack confidence in their ability every two years (23%) or three years (22%). strategy leader can help expand and shape
to execute. In Deloitte’s 2020 CSO Survey,2 most the vision of executive leaders and board
respondents (70%) rated disruptive growth as Ultimately, strategists should collaborate with members. As one executive from a leading
critical for their companies’ success, but only tech leaders to confirm that the organization’s oil and gas company said, “The CSO needs
13% are confident that their company can critical technologies support the organizational to challenge long-held views and get our
deliver on this strategic priority. strategy—and that the organization’s fellow executives to think about a market
technologists have the right framework and environment that is different from the existing
execution is critical. Agile strategy development and execution Tech-savvy C-suite. C-suite executives
doesn’t happen in a vacuum—to generate and board members should have a broad
Effective execution requires continually effective results, organizations need understanding of the critical technologies in
monitoring the underlying strategic choices foundational elements in place. which the company is, or should be, investing
and assumptions, adjusting as needed. Even to gain competitive advantage and to build
though most survey participants say they are Empowered strategy function. Whether resilience against disruption. Leaders need
seeking disruptive growth, 71% report spending it’s the CEO, CSO, or other executive, an to be supportive of investing in a portfolio
more than three months on a single round of empowered executive strategy leader is of technology investments, from proofs of
1 2 3 4 5 6 7 8 9 Strategy, engineered 16
concept to test emerging technologies to bets require the active participation of the CEO/ answers are encouraged and cataloged so they
major implementations of proven platforms. CSO to ensure that the organization’s tolerance can be used later to flag leading indicators of
Moreover, executives need to help challenge for risk is honored and well managed. the strategy’s success or failure.
the critical assumptions of those implementing
the technology to help make sure the Aligned technology and partners. Agile funding. Executing on agile strategies
implementation will be driving value for Effective organizations choose their technology demands a flexible process for planning and
the enterprise. platforms and ecosystem partners carefully, funding the technology investments required
aligning their choices and implementation to implement strategies. For insights into how
Business-savvy tech leaders. Likewise, IT decisions with their strategic goals. When selecting to finance innovation at the speed of agile, see
leaders and technologists should be engaged in important ecosystem partners, evaluate their Tech Trends 2020: Finance and the future of IT.
strategy development processes and education long-term motives and agendas to understand
that gives them a broad understanding of whether their objectives and aspirations align
the business and its strategic objectives. with yours. In a worst-case scenario, a platform Next up: Tech-enabled strategy
As strategic partners, tech leaders can help partner could become a competitive threat after
strategy and business leaders identify and they “learn” your industry. To compete in an increasingly complex world,
explore emerging technologies that support the not only are the strategies becoming more
strategic vision, with the objective of aligning Collaborative list of strategic assumptions. digital—so is the process of developing and
the organization’s technology and corporate Early in the strategy development process, lead executing strategies and monitoring outcomes.
strategies. In fact, in a Deloitte–Wall Street strategists, tech leaders, and ecosystem partners As leaders are forced to consider a wider range
Journal Intelligence survey, 40% of CEOs said should explore and challenge the assumptions of variables and future scenarios, tech-enabled
their CIO or tech leader will be the key driver of for a tech-enabled strategy to be effective. strategy platforms can help strategists think
business strategy—more than the CFO, COO, Consider holding the discussion in a neutral more expansively and precisely about the wide
and CMO combined.3 Further, big technology environment such as a workshop, where egoless range of future possibilities.
1 2 3 4 5 6 7 8 9 Strategy, engineered 17
Technology can also help leaders gain insight Finally, technology plays an increasingly development and execution. Leading
into seemingly unrelated occurrences that can important role in helping companies translate organizations are engineering their strategic
drive smarter strategic choices on a continual a strategy into strategic initiatives, facilitate function to be more agile, scalable, and
basis. Strategists should evaluate technologies strategy execution, and monitor a strategy’s stable, giving them an array of strategic
that help empower their imagination by impact in the marketplace. options in their back pocket for whatever the
identifying driving forces, informing strategic future holds.
decisions, and monitoring outcomes. (See
figure, “Strategy support technologies” on The way forward
page 18.)
Strategy development is not a one-and-done
exercise but, rather, an ongoing, cyclical
40% of CEOs said their process. As technology and business become
Trend-sensing technologies • External: Changes in society, technology, politics, economy, The US Department of Homeland Security
Identify continually scan the environment, environment, and regulations uses automated tools that scan news
collecting, analyzing, and clustering • Business: Shifts in industry dynamics, customer expectations, articles, patent filings, and more to identify
strategic leading indicators that challenge or competitor moves, and stakeholder behaviors, attitudes, and emotions emerging technologies that might disrupt
forces confirm strategic assumptions to • Organizational: Trends in products/services, operations, asset base, or improve the process of securing the
facilitate quick responses. costs, and branding United States.
Dynamic scenario tools and • Create alternate long-term scenarios to identify threats and Three health care organizations (provider,
simulators help leaders identify opportunities plan, plan/provider) developed an agent-
Inform
threats and opportunities, quickly • Develop simulations to test various strategic choices based model to simulate how competitive
strategic test potential outcomes of strategic • Enhance human understanding and insight to determine potential value-driven care dynamics could evolve,
choices choices, and prioritize critical implications, critical unknowns, and strategic responses giving each organization new insights into
uncertainties. where to play and how to win.
Analytics technologies continually track • Are we honoring our strategic choices? Deloitte uses AI-based sensing to monitor the
Monitor internal and external outcomes to inform • What’s working? What’s not working? What adjustments are needed? trajectory of two important uncertainties that
execution ongoing leadership discussions and • Is our strategy and its associated commitments generating the affect the firm’s strategy: degree of regulation
decisions regarding the organization’s projected value? and technological/automation impact on
and outcomes performance against the strategy. professional services.
1 2 3 4 5 6 7 8 9 Strategy, engineered 19
Harvard Business School should know exactly what assumptions they are making about their
strategy and the supporting technology—and the competitive results
they expect.
Maintenance scheduling? Route assignments? And just as important: What assumptions As assumptions multiply and tangle, the
Once implemented, these technology about the technology itself must be true for these human brain needs help to anticipate
decisions are baked into everyday operations results to materialize? Too often, leaders and outcomes and make course corrections.
and are difficult to change. Until recently, it’s technologists succumb to magical thinking— New technology tools can help strategists
likely that the airlines assumed that business if we invest in this technology, it will solve experiment with strategic theories, test the
travel would remain relatively stable. As all these other problems. Instead, effective assumptions’ validity and likely outcomes.
we’ve seen this year, when a key assumption strategic leaders don’t accept the promised Rapid simulation and experimentation can
turns out to be wrong, the strategy can benefits at face value—they carefully probe provide a quick, early read on possible
quickly unravel. the logic behind the proposals, searching outcomes while avoiding implementation risk.
for disconfirming data, to make sure the During execution, technologies can
Of course, the outcome of any important technology team has carefully thought through continually monitor the assumptions’ validity
strategic choice is inherently unknowable. their assumptions about how the technology in real time, providing insights that strategists
You’re never going to reduce the gray area to will advance the strategy without taking on can use to quickly and confidently course
zero, but you can install controls that warn inadvertent risk. correct if needed. Additionally, sensing
you that an important assumption is under technologies can function as an early warning
question or, in fact, invalid. One way to do this system to alert strategists when assumptions
is to ask the right questions. No one expects Everyone has access to similar are being challenged.
CEOs and CSOs to keep up with all the latest
technologies, but they should know enough to technologies—but it’s how Many of these strategy support technologies
ask the hard questions: How will this technology incorporate cognitive AI, which helps take
interact with our strategic assumptions? What you use them that makes the the human ego out of the equation when
problem will it solve, or what uncertainty will be decisions don’t work out as anticipated.
reduced? What advantage will we gain from it? competitive difference. It’s human nature to overlook or discount
1 2 3 4 5 6 7 8 9 Strategy, engineered 21
MY TAKE
shortened our planning horizon at Salesforce during this transition. But we didn’t provide the tool alerts local health organizations to
to nine-month scenarios to inform our specific actions they should take—instead, we give them time to prepare.
strategic choices about the technology that will suggested the questions they should be asking
be needed in this new world and how we could themselves in this new environment.
tackle even bigger problems. It’s up to leaders to figure out
a new future—otherwise, it
Moreover, to succeed in an accelerating world
We are also working to harness the strategic where strategies and tactics are merging,
thinking of our 50,000-plus employees. We’ve
pivoted from the centralized communications
leaders need to know what’s going on. The
ability to capture and display information for will be determined for them.
cascade we used before the pandemic to decision-making becomes hyperrelevant when
a hyperdistributed enablement of strategic you need to move this fast. Quickly adjusting Another example was the design and launch
thinking. We gave every employee a copy of strategy and tactics requires real-time data of Work.com in early May 2020. This new
our near-term scenarios and asked them to and sophisticated tools to create scenarios product helps businesses and communities
consider: How do you need to operate and adapt and forecasts. get back to work safely. We accomplished in
to this environment? How are you going to help one month what would normally take a year.
your customers do this, too? For example, we’re helping health agencies in When you move this quickly, the organization’s
California and Hawaii, as well as our own global operational cadence must accelerate as well.
Of course, we don’t send them off without organization, anticipate where the COVID-19 During every good scenario exercise, leaders
context. Every Wednesday, we have an all- virus could spread next. We’re using a mobile identify the leading indicators of potential
hands call to keep our people up to date. We survey tool that follows cell phone movement problems. Our senior executives, especially
gave them a playbook that provides guidance without capturing personally identifiable Marc Benioff, are very familiar with these
for how to think about running their part of the information. When many cell phones move indicators and frequently question operational
company and how to interact with customers from a coronavirus hot spot to another area, leaders to head off problems before they
1 2 3 4 5 6 7 8 9 Strategy, engineered 24
EXECUTIVE PERSPECTIVES
STRATEGY // The playbook for using FINANCE // Finance leaders are playing RISK // The COVID-19 pandemic has
technology to create competitive a larger role in strategy, with some even forced many companies to dramatically
advantage is changing from, “Who owns wearing both lead strategist and CFO change their strategies or adopt more
the best tech?” to, “Who uses technology the best?” hats. These leaders are drawing a fine line with dynamic approaches to the annual planning cycle.
As a result, CEOs are realizing that their collaboration technology-enabled strategy by making it dynamic Chief risk officers (CROs) should consider updating
with IT needs to go a level deeper to understand without pivoting constantly. As capabilities such as their organization’s risk profiles around key assets
whether and how their tech is providing an market sensing and constant monitoring point to (cyber, brand, core technology systems, etc.) to
advantage. Strategy executives should find the right new directions, CFOs can actively manage the capital match changing strategies. For example, companies
balance of education, control, and delegation when allocation available for such experimentation. This that relied heavily on brick-and-mortar business
making decisions about technology adoption or means placing some limits on the options available models require new risk profiles as they shift more to
implementation. In understanding what’s possible to the CEO, while helping to identify strategies that e-commerce. Going forward, strategy may become
within the bounds of their organization’s tech can generate robust profit streams in the future. even more agile and dynamic, bolstered by market
stack, CEOs can develop an overall strategy that Ultimately, the finance team may be responsible for sensing and new technology capabilities. Moreover,
plays to their strengths or allows time for needed communicating its decisions to the public in a way an organization’s demonstrated ability to manage
investments. As strategy becomes more dynamic that emphasizes purposeful experimentation and risk amid uncertainty and disruption could become
in the coming months, CEOs would be wise to profit potential, especially at a time when analysts and a strategic differentiator. CROs can decide which
remember that their responses to trends will be only investors are highly attuned to uncertainty. kind of leader they will be: one who manages risk
as good as their organizational technology permits. by using governance to slow down adoption of new
technologies, or one who optimizes risk in new
technologies to deliver stronger business outcomes.
1 2 3 4 5 6 7 8 9 Strategy, engineered 26
1
Does your technology limit your Learn how corporate strategy
organization’s strategic options? If leaders are evolving to navigate
so, how can that be addressed? uncertainty and change strategic
planning.
2
work better to understand the between digital maturity and
strategic plan as well as the financial performance
opportunities and constraints of Gain insights into the key “digital pivots”
your technology architecture? that can affect financial performance.
3
for your strategy? How do you See how lessons of digital
monitor these and make timely transformation can prepare an
adjustments when needed? organization’s response to sudden
and severe disruptions.
1 2 3 4 5 6 7 8 9 Strategy, engineered 27
AUTHORS
Our insights can help you take advantage of emerging trends. SENIOR CON T R IBU TOR S
If you’re looking for fresh ideas to address your challenges,
let’s talk. Nicholas Reed
Senior manager,
Rich Nanda Deloitte MCS Limited
US Strategy offering leader
Bruno Kakoobhai
Deloitte Consulting LLP
Manager,
rnanda@deloitte.com
Deloitte & Associados, SROC S.A.
ENDNOTES
1. Gerald C. Kane et al., The Transformation Myth:
Leading Your Organization through Uncertain
Times (MIT Press, forthcoming 2021).
Core revival
TREND 2
Core revival
New technologies, techniques, and business cases to drive your modernization strategy
S
ince Tech Trends debuted 12 years (ERP) systems by reducing technical debt and initiatives. Likewise, they want more from
ago, we have explored many of the leveraging non-ERP platform capabilities in a these investments than mere improvements
transformational approaches and more manageable, cost-effective way. to the enterprise IT plumbing—they want to
technologies that CIOs may take to extract create a lasting foundation for innovation and
more value from legacy core systems. This In the current economic climate, it’s more competitive advantage.
year, in what we recognize as an exciting strategically important than ever to help
new trend, some CIOs are exploring several your legacy core systems support the agility, Over the next 18 to 24 months, we expect to
innovative ways to redefine the core innovation, and new modes of working see trend participants:
modernization business case. For example, that fuel that digital potential. Yet costs
new techniques are making efforts to lift- associated with popular approaches to core • Reengineer costs, project funding models,
and-shift to the cloud more cost-neutral and modernization—particularly those used and the desired outcomes associated with
operationally manageable. Likewise, an array with custom-coded business transaction their core (and, potentially, data center)
of leading-edge technologies—including systems such as application rewrites modernization use cases by focusing on
low-code/no-code, more intelligent business and cloud migrations—can give some third-party platform management services.
rule discovery, and core mapping—offer new organizations pause.1 Faced with pandemic- These services feature nontraditional
ways to revitalize valuable core assets. Finally, related uncertainty and tight budgets, funding arrangements and operate-to-
looking beyond lift-and-shift, organizations many IT leaders and their C-suite peers are transform approaches that combine
can upgrade enterprise resource planning looking for new ways to fund modernization modernization and innovation investments.
1 2 3 4 5 6 7 8 9 Core revival 31
Trend participants can also lift-and-shift big, Then, going forward, CIOs and their teams that in this time of uncertainty need an
custom-coded transactional systems to will make key decisions about whether to efficient, cost-effective way to move rigid yet
hyperscaler cloud platforms. develop mission-critical capabilities in the essential core assets. Revitalized in the cloud,
ERP system itself or on alternate platforms. these assets can provide a strong foundation
• Explore opportunities to accelerate the for mission-critical innovation and growth
discovery of the internals of black-box strategies in areas such as artificial intelligence
legacy systems to facilitate modernization. Lift, shift, and thrift (AI), edge computing, and quantum.
In addition, leaders will explore ways to use
technologies such as PaaS low-code and During the past few years, some companies
no-code platforms to perform traditionally began looking beyond lift-and-shift for Making a fresh business case
complex tasks and integrations via point- opportunities to redesign core capabilities
and-click, rather than by writing new code. in the cloud;2 they wanted to take full In the coming months, we expect to
advantage of cloud platform offerings to see creative approaches for financially
• Support a platform-first strategy by enhance their existing capabilities. But many reengineering the core modernization business
deploying leading-edge system analysis leaders were wary of potentially large future case gain traction in the marketplace. Though
tools to identify redundant or extraneous costs.3 Even before COVID-19 disrupted the specifics will vary by need and industry,
code within legacy ERP systems, and either the global economy, pragmatism—and a their common ground will be that they offer
move this code to another platform or growing focus on the business case for core organizations the following ways to achieve
delete it altogether. The ultimate goal of modernization—had begun informing more their core modernization goals.
this approach is to remove technical debt transformation decisions.
that accrued over years of using ERP as a • Operate and transform. Organizations
development environment and, in doing so, Today, we see renewed interest in migrations are exploring other creative operate-to-
return ERP assets back to a baseline state. to the cloud, particularly from organizations transform agreements with implementation
1 2 3 4 5 6 7 8 9 Core revival 32
partners, holding them accountable for have sprung up over time to work around over time. Professional services firms are willing
migrating and/or upgrading systems, other limitations, leaving burdens of to invest in back-loaded arrangements, to be
and utilizing powerful tools. While these technical debt, outdated applications, and rewarded for long-term impact while easing
agreements will inevitably vary in detail and workarounds. Transitioning collections immediate cash-flow burdens. In some cases,
scope, many arrangements are designed of systems to the cloud may enable all up-front fees can be deferred in exchange
to minimize capital-intensive efforts. In (or force) the long-overdue process for a share on the upside of the business case.
fact, some organizations are able to secure of rationalizing redundant systems, For clients, these options offer fast access
agreements whereby their systems will be eliminating unnecessary dependencies, to needed platform capabilities on what are,
modernized to cloud-native platforms in a and modernizing capabilities. Modernizing for some, attractive terms. The hyperscalers
few years, while organizational operating a collection of related systems—or retiring themselves can get more core systems
expenses stay neutral. some of them altogether—can lower care- operating in their clouds, make those clouds
and-feeding costs, increase efficiencies, increasingly sticky, and over time support
• More bang for fewer bucks. During and enhance system performance, all of additional workloads for their customers.
the last few years, vendors have made which may lower cost allocations across
significant improvements in proprietary the stack and bolster your business case
tools that support transition to the cloud. for a core revival initiative. Transforming custom code
By simplifying the process, these tools are
giving rise to some compelling business Building a fresh business case for core New and improved technologies can help
cases in which migrations can be cost- modernization represents a different kind organizations revitalize legacy systems to
neutral or lead to cost savings. of transformation play. Hyperscalers are either spruce up or retire core systems. Using
increasingly willing to chip in funding to help the following advanced techniques, they can
• System rationalization—how less is organizations transition to cloud offerings carve out pieces or wholesale replace
more. In many organizations, systems with expectation of recouping their investments legacy systems:
1 2 3 4 5 6 7 8 9 Core revival 33
• Improved low-code platforms. Low- on tech leaders and developers. Grand the application itself with a microservice.
code (LC) platforms are dramatically View Research valued the global low- In the context of core modernization, this
more capable than even a few years code application development market at represents a game-changing breakthrough:
ago. Some enterprise technologists may US$11.45 billion in 2019, and expects the Not only can existing business logic fuel
harbor doubts about LC scalability and market to grow 22.7% annually until 2027.4 requirements or LC development—the
performance, but the list of industry- tools are improving as tool developers apply
specific LC platform opportunities • Modernized business rule extraction. AI/ML to increasingly automate aspects of
continues to grow. Using LC offerings Traditionally, the process of identifying the code extraction process.
from Appian, OutSystems, Salesforce, and extracting business rules from
ServiceNow, or other vendors, systems custom code in legacy applications has • Improved incremental modernization.
designers can carry out complex tasks and required an extensive, manual effort Another novel approach to application
integrations through point-and-click rather and an army of specialized resources. modernization involves a process known
than having to write code. Moreover, LC Today, improved mining technologies as core mapping. Using a set of increasingly
vendors are racing to integrate advanced and approaches make it possible to sophisticated mapping tools, legacy
AI/machine learning (ML) capabilities into peer inside legacy code—regardless of systems can be visualized as a connected
their platforms to help augment the user language—and extract its business logic graph of constituent parts. A map of
experience. Finally, “case” and contact with less effort and higher fidelity. By connected code modules works similarly
management are a core function of many scanning the code of an application that to a social network graph on Facebook, in
of the LC platforms—and often an area of is a candidate for modernization, you can which lines and dots represent connected
great need for agility in legacy platforms. quickly identify essential business logic, as people. By identifying logical subgroupings,
As such, the impact that LC platforms well as hot spots where a system issue is system engineers can identify and sever
can have on core modernization efforts taking place, and either refactor the code legacy interfaces, replacing the interfaces
cannot be overstated, a fact not lost in question, remove it, or perhaps replace with modern API and service-based
1 2 3 4 5 6 7 8 9 Core revival 34
techniques. Over time, these services can priorities of yesteryear. Addressing these significant long-term investments in ERP
then be modernized individually as needed challenges can have the net effect on your ERP assets have begun questioning the wisdom
in a more predictable, consistent way. systems of jettisoning unneeded ballast. of spending seven or eight figures on a
major ERP upgrade. Will these expensive,
Building a fresh
complex undertakings deliver lasting
Breathing new life into legacy ERP systems material benefit to the organization? Will
LESSONS
how to reduce their existing technical debt. Upon finalizing a
Albemarle takes on the core detailed consolidation plan, the company kicked off its business
FROM THE and more transformation journey in mid-2017. During the next 18 months,
the Albemarle team executed a series of four ERP deployments,
FRONT LINES Specialty chemical company Albemarle has completed its resulting in a single ERP platform that serves the entire global
enterprise with common processes and best practices. Along
journey to one consolidated ERP system—a journey that in many
ways exemplifies the art of the possible for core modernization. the way, Thompson and his IT team migrated the global
operation from Lotus Notes to MS Office 365. (This latter step
After a series of multibillion-dollar acquisitions, this global is no small feat in an organization with operations ranging
leader of lithium, bromine, and catalyst solutions found itself from remote mining locations in Chile and Western Australia to
with a stable of far-flung business units. Albemarle needed to manufacturing facilities in the Americas and China.)6
integrate these acquired units, create a shared service center,
and establish common best practices as soon as possible. Albemarle’s next step is notable: Once Thompson and his
But a major obstacle stood in the way: The business units in team had everyone operating on a single ERP platform,
question did not share a common ERP platform. “Creating a they executed a “big bang” version upgrade to the global
single ERP platform was going to be essential to that business ERP instance in a mere six months. How did this group of
integration initiative,” says Albemarle CIO Patrick Thompson. technologists achieve what may be a speed record for ERP
“Not only would it support our shared services model—it upgrades? First, they eliminated their technical debt by
would eliminate the technical debt from more than 3,500 electing to make no code changes or customizations to the
customizations that had been made to the two platforms.”5 global instance they deployed—it was basically out of the box,
dramatically simplifying software upgrades. “We did create
Albemarle spent almost a year vetting the business case 70 extensions, but they were outside the code, and we used
for consolidation, refining the ROI metrics, and aligning on APIs for those,” Thompson says. “We also had really good
1 2 3 4 5 6 7 8 9 Core revival 37
methodology and talent. Over the course of digital enhancements like these possible,” But when, in 2017, the company found itself
this project, we have put a lot of work into Thompson says. “We no longer have to spend with a burning legacy platform that could
developing the foundational framework and money on technical debt and customizations. no longer support its needs, Sogrape head
the talent that we need to execute upgrades. Instead, we can invest those resources to make of IT Carlos Alves began looking for a way to
Going forward, we will be able to flex this leapfrog progress with digital technologies that harness core modernization in the service of
muscle up and down as needed for upgrades.” can transform our company.” methodical, artisanal winemaking. “We needed
a consolidated platform that could connect
As of January 2020, Thompson had completed
the consolidation and upgrades, and began
Centuries-old tradition all the functional steps in our winemaking
process,” he says. “At the same time, we—as
shifting his focus toward broad digital meets a low-code future a global organization operating in the 21st
transformation. The company is accelerating century—needed speed and agility.”8
its work on IT/OT convergence, particularly In a family-owned company like Sogrape
in its manufacturing operations. In the back that embodies the painstaking traditions Alves considered upgrading Sogrape’s in-
office, Albemarle will be using robotic process of Portuguese winemaking, the immediate house core systems to cloud-based offerings
automation and business process mining benefits of modern technology’s speed and by the same vendors but instead chose to go
tools to elevate its first pass yield and rework agility aren’t easy to see—after all, the journey with the OutSystems low-code platform. “We
process performance. And, on the customer from vineyard to bottle can never be rushed. considered three factors: time to delivery, cost,
side, the company is exploring opportunities Consumers in more than 120 countries prize and training time required to use the new
to extend to customers some self-service Sogrape’s traditional approach to winemaking; technologies,” he says. “In each area, low-code
ordering and track-and-trace. over the years, public figures such as Jimi worked best for us.”
Hendrix, Queen Elizabeth, and Steve Jobs have
“The work we did consolidating and savored the vintner’s unique signature brand, Using OutSystems low-code technology, Alves
modernizing our foundational platforms makes Mateus Rosé.7 and his team built “Wine Connection,” an
1 2 3 4 5 6 7 8 9 Core revival 38
integrated platform that offers on-site process developed a formal Agile methodology for low- Understanding the precise density of specific
control and real-time information-sharing to code projects. Using this template, Alves and harvest batches can help winemakers
functional groups across Portugal, including his team are standardizing the process that control the fermentation process.
viticulture, wineries, bottling, registration, departments across the enterprise will use to
contracts, and others. Users can access the create new products. “For example,” he says, Says Alves, “Wine Connection strikes the perfect
platform via desktop, tablet, or mobile phone. “we are starting a new three-phase project for balance between innovation and tradition.”
human resources to develop a platform-based
public cloud providers’ analysis of alternative energy toward migrating systems to a third- IT undertook a similar process of careful analysis
resources to make key decisions about the party platform.” Moreover, the ability to push before deciding to migrate from a legacy on-
organization’s future architecture. But in the the burden of managing some core capabilities premises database to a cloud-based alternative.
end, says Bill Livesey, GM Financial’s senior to a cloud provider was an attractive option, “This was a big decision given the size of our
vice president of digital software solutions, the particularly for an IT team that had been loan origination system and the sensitivity of the
most compelling business case called for using gradually spending more and more time financial data contained in it,” Livesey says. “We
cloud platform-as-a-service, when possible, to maintaining aging on-premises systems. Finally, ultimately became comfortable that a cloud-
modernize legacy systems already in place.10 business teams stood to benefit as well. Over based solution could meet our standards for
the course of the project, the business and IT security and privacy.”
“It came down to controlling our destiny. We would have an opportunity to forge a strong
want to maintain our competitive advantage collaborative partnership that could deliver GM Financial has more work to do as it
using core systems that we own and control,” innovation opportunities, enhanced operational reimagines its legacy loan origination platform in
Livesey explains. “We’ve invested so much of efficiency, and more frequent deployments. the cloud. But even in this first leg of the journey,
our intellectual property in these platforms for the project enjoys broad support from across
so many years, it just doesn’t make sense to During the first leg of GM Financial’s the organization. “Our partners in the business
give away that IP to others.” modernization journey, Livesey and his team are excited about this effort,” Livesey says. “We’re
went through a process of determining which taking a very large, sprawling architecture, and
The business case for cloud and PaaS also system components were candidates for transforming it into a single, consolidated loan
included cost-related elements that Livesey moving to PaaS. As it turns out, many were origination platform. They get powerful, reliable
could not ignore. “With PaaS, we could keep “very suitable” and will be migrated with few tools to support their work, and IT will get a
developing the products and services our changes in the near future. Others, due to age stable, manageable production environment
business partners need right now. We wouldn’t or complexity, had no path to the cloud and will that we can modernize on an ongoing basis with
have stopped everything and shifted all of our have to be refactored or deleted altogether. minimal effort. Everybody wins.”
1 2 3 4 5 6 7 8 9 Core revival 40
MY TAKE Over the last few years, I’ve noticed a big shift
Justin Kershaw in the way boards and executive leadership
think about enterprise technology—a
Corporate vice president shift that is redefining the way companies
and CIO, Cargill approach core revival initiatives.
Many of these decision-makers have traditionally viewed technology
primarily in terms of cost, as in: “How much will this new system cost
right now?” But many of these leaders are becoming more forward-
looking, focusing less on technology’s cost and more on its long-term
value. This is a welcome change.
At Cargill, we have been on a core important not just to business success. In For example, we developed a single, unified
modernization journey for several years. From many cases, it’s vital to the safety and well- platform called Maestro for Cargill’s strategic
our HR systems to treasury, from distribution being of our societies. sourcing department, an internal operation
and transportation systems to manufacturing, that accounts for more than US$5 billion
we have modernized more than half of Cargill’s The modernization journey for Cargill began in indirect spend annually. The investment
systems. We have also invested heavily to with a shift to third-party managed services, in Maestro—a foundational platform that
update our approach to technology risk, both enabling us to reinvest in core networks and replaced dozens of systems—has modernized
internally and externally. Such efforts are no infrastructure. We invested in technologies like our sourcing operation. We’ve achieved
small feat for a 155-year-old company with advanced analytics, cloud, and ERP systems. significant sourcing improvements over
operations in 70 countries and an ambitious Our leadership team and board are now seeing the last few years that are improving the
purpose of nourishing the world in a safe, the value and returns from these efforts. company’s overall results.
responsible, and sustainable way.
Going forward, we will move away from talking In the analytics space, we stood up a data
For a company like ours, maintaining and about “applications” and “infrastructure.” platform a few years ago and provided our
operating core technology is more than just Instead, the next phase of our modernization employees access to two self-service analytics
good business. Consumers and farmers, is creating “platforms” and “services.” This will tools. We went from basically having no
as well as some of the world’s biggest food ultimately change our talent requirements— centralized, shared data to a platform that
companies, retailers, and restaurant chains we will need more people who can build and handles some 2 million daily transactions.
depend on Cargill. The pandemic-related maintain platforms and services. It will also Our data platform lets us wield information at
disruption of supply chains and the global change the way we will work with third-party the speed and scale our business demands,
economy in 2020 served as a potent reminder technology providers, and how we serve allowing our teams to make better, data-
to Cargill and companies everywhere that our internal business partners and Cargill’s driven decisions. Meanwhile, our self-serve
operating modern technology is critically external customers. analytics tools and coaching services are now
1 2 3 4 5 6 7 8 9 Core revival 42
helping about 30,000 employees. But the platforms, services, and digital products to meet And even beyond modernizing, executives are
most impressive number—especially to our those needs. Soon, Cargill customers using their coming to understand that progress on these
executive team and board of directors—is the phones will be able to access an entire portfolio technology journeys is vital to an entirely new
more than 7x return on these investments. of digital services, some even connecting them form of disruptive competition in the market.
directly to the manufacturing floor. In IT, we will Today’s savviest business leaders know that
We are also investing heavily to modernize still have engineers with traditional skill sets the future belongs to companies that put in
our processes. Like many global corporations, focused on maintaining and modernizing our the hard work to future-proof their technology
we have some inefficient, disconnected technology stack. And we’ll also have teams foundation.
systems—and more than a few manual tasks focused on digital platforms and services.
that can be automated. We are tackling this Together, these teams will build the kind of
head-on. Working with our ERP vendors, we modernized platforms and processes that we Rather than evaluating a
are creating perpetual, end-to-end processes
throughout our organization. This is a complex
need to support a digital ecosystem.
technology investment for
task that involves system upgrades, process
engineering, and a huge change-management
Modernizing our core systems and
augmenting them with new services and
its costs, these leaders view
effort. Putting new technologies in place is one
thing—getting people to use them effectively
platforms is transformational—and hard—
work. But in the current technology and
these investments as continual
is quite another. That said, we are making economic climate, hard is relative. Savvy modernization that will create
enormous progress on both fronts, and it has C-level leaders and boards of directors now
been well worth the effort. get it: Rather than evaluating a technology positive returns.
investment for its costs, these leaders view
My team’s approach in the future will be these investments as continual modernization
to identify business needs and then build that will create positive returns.
1 2 3 4 5 6 7 8 9 Core revival 43
EXECUTIVE PERSPECTIVES
STRATEGY // Ongoing investments FINANCE // Many CFOs may have to RISK // Many core modernization
in core systems are necessary for IT to make a crucial strategic decision about the initiatives today introduce leading-edge
be an enabler of business value. Too future of their organization’s core systems: technologies into existing IT ecosystems.
often, executives view such investments in terms procrastination or proactive modernization? The larger Risk officers should expect risks to emerge when
of immediate cost rather than future value. This is and more complex an entity is, the harder—and more old and new technologies intersect; they can work
shortsighted: In many digital nonnative organizations, expensive—an enterprise-level project such as core closely with CIOs to anticipate and manage risk in a
legacy core systems provide essential foundations revival can be. Some CFOs may choose to wait for manner that doesn’t impede modernization efforts.
for critical digital and business transformation existing systems to break or become untenable before Likewise, they can collaborate with the business side
initiatives. Several nontraditional approaches for acting, but they can accumulate technical debt along to identify opportunities to leverage risk management
financing core projects can help cost-focused the way. By contrast, finance leaders who are eager in new tech-enabled products and services as a
CEOs find a different path to modernization. For to modernize can consider a variety of approaches strategic differentiator that helps build consumer
example, working with the CFO, they can identify beyond large-scale migrations. For example, exploring trust. With these twin objectives in mind, CROs can
more cost-friendly opportunities to partner with creative deal structures with large cloud vendors or help their organizations balance technology priorities
hyperscaler cloud vendors or major ERP vendors moving existing capabilities to low-code platforms can and consumer sentiment against risks and regulatory
and pursue smaller revitalization projects using low- lead to cost-neutral options for establishing a future- concerns.
code platforms. These and similar approaches can ready foundation within core systems. Leveraging
fundamentally rework the traditional business case these and other core revival approaches, CFOs
for modernization by transforming a line-item cost can become catalysts for fiscally responsible digital
into an investment in the company’s future. transformation and avoid being painted as a financial
hurdle to the IT department’s big goals.
1 2 3 4 5 6 7 8 9 Core revival 44
1
Where would negotiating operate- Stay on top of the latest
to-transform arrangements with cloud news, views, and
your technology vendor(s) be real-world insights.
most useful?
2
Would you benefit from moving Gain insights on the benefits of
legacy applications to more low-code platforms and how it can
modern platforms (for example, enable greater agility to respond to
low-code or cloud options)? business needs.
3
What is your strategy for Explore a strategic approach to core
eliminating technical debt in your modernization through clean ERP
legacy ERP system? and reduced technical debt.
1 2 3 4 5 6 7 8 9 Core revival 45
AUTHORS
Our insights can help you take advantage of emerging trends. SENIOR CON T R IBU TOR S
If you’re looking for fresh ideas to address your challenges,
let’s talk. Thorsten Bernecker Franscisca Wahjudi Matt Fisher
Principal, Managing director, Specialist leader,
Scott Buchholz Deloitte Consulting LLP Deloitte Consulting LLP Deloitte Consulting LLP
Government & Public Services chief technology officer
Darwin Deano Miguel Amaro Anantha Ramadas
Deloitte Consulting LLP
Principal, Associate partner, Senior manager,
sbuchholz@deloitte.com
Deloitte Consulting LLP Deloitte & Associados, SROC S.A. Deloitte Consulting LLP
ENDNOTES
1. Angus Loten, “Lift-and-shift cloud strategies can 9. GM Financial, “About us,” accessed October 7, 2020.
be costly,” Deloitte CIO Journal on the Wall Street 10. Bill Livesey (senior vice president of digital software
Journal, December 18, 2018. solutions, GM Financial), phone interview with
2. Karl Schwirz and Michael Hodgdon, “Is lift & shift authors, September 14, 2020.
actually a quick and painless path to the cloud?,”
Slalom, March 2, 2017.
Data becomes even more valuable when RPA, cobots, and drones make supply chain
it is aggregated, mined for insights, and field work safer and more efficient.
disseminated in real time across an
ecosystem of stakeholders.
1 2 3 4 5 6 7 8 9 Supply unchained 48
TREND 3
Supply unchained
Transforming a traditional cost center into a value driver
O
ver the next 18 to 24 months, customer needs and to fluctuations While companies have been digitizing their
we expect to see manufacturers, in demand. supply chains for some time, recent shocks
retailers, and others take supply have provided fresh urgency for change. In
chain transformation to the next level by • They can explore ways to capture larger Q1 2020, the world witnessed a decades-old
optimizing their supply chain ecosystems volumes of structured and unstructured supply chain model being disrupted, and it
for resilience and risk. Moreover, they will data. By mining this data for operational wasn’t pretty. Driven by panic buying, consumer
begin transforming their supply chains from insights, they can continuously optimize demand for paper products, cleaning supplies,
traditional back-office cost centers into value- systems and processes throughout the and other nonperishable goods spiked
driving operations. How? By following industry organization. And by sharing the data dramatically.1 Yet within domestic and global
leaders’ playbook for optimizing supply chains more widely, they can look to optimize supply chains, these demand signals were not
for customer segments: their entire supply ecosystem. transmitted quickly to manufacturing floors
and sourcing departments. After years of small-
• Trend participants can deploy an array of • Organizations may also pursue scope efforts to optimize and strengthen their
digital tools to hypersegment customers opportunities to use robots, drones, and supply networks, many leaders were confident
and capture demand signals from other technologies to make supply chain that they had created robust, transparent
disparate parts of their value chains. They teams more effective, efficient, productive, supply networks. When COVID-19 hit, it quickly
can then use this information to make and safe. became clear that their efforts had been
supply chains more responsive to unique insufficient. Even the most flexible, resilient
1 2 3 4 5 6 7 8 9 Supply unchained 49
supply chains proved to be only as strong as supply chain management and procurement.4 of globalization, interdependency, and—
the weakest links of their suppliers’ chains. As a result, some are now taking a page importantly—vulnerability.
from digital supply network pioneers. Using
The ensuing disruption was widespread. In a analytics to understand their customers Is your supply chain ready?
recent survey of supply chain professionals, more deeply and predict their behavior,
97% of respondents said their organizations a growing number of organizations are
experienced COVID-19–related disruption.2 working to understand the sources of value From cost center to value driver
Some of these organizations were caught off for and buying behavior of those customers.
guard by the severity of the pandemic and By sharing information with their network The idea of transforming the supply chain
the speed at which it unfolded. They shouldn’t of suppliers—and receiving information from cost center to value driver is not new.
have been. The world has become ever in return—they can better link supply and Over the past two decades, leading companies
more dependent on connectivity, vulnerable demand. And understanding their customers’ have fine-tuned strategies for optimizing
systems, and global supply chains, and this value drivers makes it possible to optimize for incentives and disincentives for online
kind of high-impact disruptive event will likely deployment, logistics, procurement, and more purchasing and delivery timing—strategies
occur with greater frequency.3 in their new supply networks. that manufacturing, retail, and other sectors
may find helpful as they transform their supply
Against this backdrop, a growing number of The work of turning a traditional cost chains. (See Lessons from the front lines
leaders are realizing that their organization’s center into a robust value driver will not “Pactiv Evergreen gets proactive with factory
supply chain optimizations may not suffice happen overnight. Nor will building the asset intelligence,” on page 55).
in an interdependent, unpredictable flexibility, transparency, and resilience that
world. Indeed, in the same survey, 73% of complex supply networks need to weather Online retailers were among those
respondents said their organizations are now disruption. But the time to start is now. pioneering the art of using predictive models
planning major shifts in the way they approach COVID-19 portends a new, uncharted phase to optimize the location and volume of
1 2 3 4 5 6 7 8 9 Supply unchained 50
inventory, procurement, and replenishment. priorities helped these organizations their impact on the supply chain, from point
Using customer data, they developed highly pre-position products closer to demand, of sale to the manufacturing floor and all the
detailed customer cost-to-serve profiles decrease transit time and risk, and increase way back to tier-three suppliers. Indeed, the
used to segment customers into groups delivery schedule reliability. Meanwhile,
5
extent to which customer information can be
based on location, preferences, and service they could maintain remote warehouses captured in real time to feed supply chain and
expectations. These retailers found that in to supply nonurgent deliveries to the more manufacturing production decisions is already
some cases, customers will pay premium price-sensitive. These insights transform becoming a competitive differentiator.
prices for premium delivery services, while supply chains into something new: a tool that
more price-sensitive customers will accept encourages customers to make informed, Take, for example, a consumer products
longer delivery time frames. personal buying decisions while simultaneously company that manufactures and sells liquid
improving company profitability. laundry detergent in plastic bottles. Analysis
tool for engaging customers. By optimizing organizations understand the value they from these and other supply chain
the component parts of its supply chain to provide customers, develop greater clarity into functions into a commonly shared
differentiate services provided to each customer, internal operations, and work to make supply data platform enhances end-to-end
the company can find and maintain the sweet more visible across their networks. Data— transparency, live metrics that support
spot between cost and delivery service. from internal supply chain operations and human and machine-based decision-
external partnerships alike—is the keystone making, and operational efficiency.6 In
Tightly controlled, robust supply networks can for these efforts. Enhanced data visibility and addition to IIoT sensors, visual, acoustic,
offer another advantage as well. When faced speedier data processing can fuel efforts to and temperature monitoring tools can
with rapid, unexpected spikes in demand such align the supply and value chains. generate unstructured and
as we saw in the early months of the COVID-19 nontraditional data streams that,
pandemic, digitized, data-driven supply chains Over the next 18 to 24 months, we expect once digitized and analyzed, can help
that provide high levels of transparency to see organizations taking part in the supply maintenance teams identify anomalies
may be able to synchronize their planning, unchained trend take the following steps to and perform predictive maintenance.
production, and fulfillment functions capture and analyze more data:
effectively and minimize—or even prevent— • Boost data capabilities at the edge.
widespread disruption. • Leverage IT/OT convergence. The same In the arena of data management, time is
smart factory applications and Industrial money. Time-sensitive data can become
this process by moving processing and offer visibility into every organization’s supply operating anywhere within a supply chain
storage capacity closer to the source chain, not just their own. The platforms may network to share information. This can help
of data. In this distributed architecture bring an AI and advanced analytics layer address a perennial challenge of one group
model, data does not have to go to the positioned on top of all the information to in the chain building a product or data model
core or cloud for processing, analysis, and enrich the entire data corpus. Data, then, that others inside an organization alone
dissemination. For example, digital data becomes interoperable. cannot easily replicate or support. There can
generated at the point of manufacture or be a second level in the data framework that
may not have their own analytics and processing can fuel
compute capabilities. New tools for supply chain teams
efforts to align the
Meanwhile, as organizations optimize their
supply and value chains.
When we think of supply chain in its historic
internal operations to serve clients and role as a cost center, we cannot overlook the
customers better, they are realizing that they cost, safety considerations, and inefficiencies
need more visibility into their external sources The journey to full data interoperability will associated with some non–value-added
of supply. Some are starting to explore the take time. As a first step, consider constructing tasks performed by supply chain talent. For
idea of creating common logistics platforms a two-tiered data framework that incorporates example, consider a traditional fulfillment
that can be used to share information across elements of a shared-data future. On one model: When an order comes in, a coordinator
all the suppliers in the network in real time. level, data will be interoperable. Companies hands a printed form to a forklift driver.
When platforms become transparent, they can create a native standard that allows users The driver goes into a warehouse, lifts the
1 2 3 4 5 6 7 8 9 Supply unchained 53
purchased product onto a palette, and then As the supply unchained trend gathers steam • Aerial drones. Companies can use
drives to an adjacent rail yard, where he in the coming months, we expect to see more unmanned drones for a variety of tasks, from
loads the palette into a boxcar. Though an organizations address this challenge head-on providing inbound logistics in time-critical
integral part of many supply chain operations, with an array of technologies: situations to carrying materials from storage
processes in which human workers operate to factory and transporting directly from
heavy machinery in transit hubs and enclosed • Autonomous robots and collaborative receiving to shipping. Drones can also scan
warehouses are often costly and inefficient. cobots. Implementing autonomous robots inventory efficiently and reduce labor costs.10
They may also carry a degree of safety risk. In can drive value by reducing direct and
the energy and utility industries, where field indirect operating costs and increasing • Computer vision. Cameras are rapidly
teams work with power lines and telecom revenue potential. They can lower labor becoming ubiquitous and connected. Supply
towers in remote locations, the risks and the costs and increase productivity by working chain operators are placing them, in tandem
costs can be even higher. around the clock.8 Likewise, cobots work with AI, throughout warehouses and freight
alongside human workers, augmenting yards to count stock. Companies are also
their performance. Their movements using these computer-vision technologies on
Organizations are are easily programmable, which enables factory floors and in offices to monitor social
Interoperable data, AI, and machine learning about globalization and business-critical
also have a role to play. The ability to tie even dependencies—it has laid bare the
the most remote supply chain functions into vulnerabilities of traditional supply chain
a seamless network with real-time data and, models operating in a world where large-scale
then, automate those functions or control disruption may be no longer the exception
them from a central location will be critical to but the rule.
lowering costs, while enhancing worker safety
and efficiency.
LESSONS
operating costs by making the company’s existing industrial
Pactiv Evergreen gets proactive assets more efficient.
Pactiv Evergreen wanted to transition from down, the entire production line goes down. OEE and a positive impact on the bottom
reactive maintenance to condition-based Using acoustic signal analysis, Pactiv Evergreen line. “The results so far have been excellent,”
monitoring, as a precursor to building up developed an algorithm based on machine says CFO Mike Ragen. “We have seen a lift of
sufficient data history to become predictive. learning that would predict grinder blade wear about 19% in output, and that that equates on
By utilizing edge processing technologies to and alert maintenance to replace the blades one line to about US$2 million of profitability.
capture critical data points such as vibration, during the next changeover, thus preventing Extrapolating that across the 18 lines, we
temperature, and pressure, leaders were unplanned downtime. should see a US$36 million lift.”11
able to predict downtime and failures based
on anomalies in real time. This information Finally, Pactiv Evergreen developed a factory
fed a library of proprietary condition- control tower to allow plant leadership to
based monitoring applications, which were monitor line asset performance as well as
customized for different user groups. the OEE for each line and the entire plant.
These apps pushed actionable insights to The control tower also provides insights on
the right people, who could then address raw material consumption (blend insights) to
potential problems before production control quality and material usage variances,
disruptions occurred. machine and human performance insights
(activity insights), and overall production
As part of the initial plant deployment, project health (production insights).
teams also experimented with signal analysis
and video analytics to understand asset Pactiv Evergreen’s factory asset intelligence
health throughout the production process. For initiative has transformed the company’s
example, the grinder is the most unintelligent entire approach to operational monitoring
asset on the production line, but if it goes and maintenance, resulting in a 9% boost in
1 2 3 4 5 6 7 8 9 Supply unchained 57
and executive director, NIAR, Cold War yet continue to serve as workhorses in America’s strategic
air wing. Their longevity can be attributed, in part, to a complex supply
My team at Wichita State University’s movement underway in which organizations In our work with smart factory technology at
National Institute for Aviation Research (NIAR) are integrating design, manufacturing and Wichita State, I see how some of the digital
collaborates with the US Air Force to develop other components within larger supply chains. advances we are making with smart factories
digital tools and predictive models for the B1-B Entire ecosystems connected by a digital thread can help transform other components in a
airframe. For example, we have engineered increasingly share the same high-quality data digital supply network. A smart factory is a
digital-twin capabilities that offer maintenance in real time from end to end. A part supplier highly digitized and connected production
teams an unprecedented 3D view into is now connected to designers and engineers facility that uses technologies such as artificial
how particular aircraft parts will stand up who are, in turn, sharing data with workers and intelligence, IoT, and robotics to manufacture
to use over time. In the coming years, this machines on the factory floor, and on through products; it can self-adapt and autonomously
capability will help aviation mechanics perform to warehousing and fulfillment. Every aspect of optimize manufacturing operations. Machines
predictive maintenance on the B1-B and the product life cycle is integrated into a unified, on a smart factory floor can automatically
other legacy aircraft when—and only when— data-driven, digital process that optimizes costs boost production based on demand signals
needed. What’s more, using digital-twin data, and efficiency over time. or slow down production based on supply
they will be able to 3D-print long-obsolete signals. Likewise, with smart factories acting as
EXECUTIVE PERSPECTIVES
STRATEGY // Transforming an FINANCE // To manage current supply RISK // Global supply chains have faced
organization’s supply chain from cost chain disruptions, and to make supply acute disruption and increased regulatory
center to profit driver can have a significant chains more resilient for the future, scrutiny as a result of the pandemic; risk
impact on overall business strategy. While CEOs CFOs can explore opportunities to invest in new leaders today may not even be aware of all the points
may not dive too deeply into the minutiae of data technologies for greater integration. Case in point: of risk in their supply chains. Even so, they may be
interoperability and demand signal capture, they blockchain.12 The CFO, along with the broader C-suite, able to respond effectively to future disruptions
should focus on the potential value that data-driven may be interested in the possibilities of increasing by learning more about vulnerabilities in their
customer segmentation, digital optimization, and blockchain adoption for greater supply chain supply chain. Technology now makes it possible for
ecosystem transparency can help create. Taken resilience in the postpandemic world. In fact, 40% organizations to understand the risk profile of their
together or individually, these opportunities are of CFOs in a recent Deloitte survey expected their vendor landscape, share data in real time, protect IP,
relevant to business and financial planning, risk supply chains to be more diversified after the effect and track contractual terms. More disruptions are
management, and organizational efficiency. Notably, of coronavirus.13 To achieve this diversification—and inevitable in increasingly complex supply networks,
they can also help organizations optimize their stay profitable—CFOs may need to dive deep into but forward-thinking CROs can take steps today to
supply chains for resilience and flexibility, which in the tactical items such as supplier payment terms, vendor reduce potential points of failure throughout all parts
current global economic environment should be on assessments, and interest rate negotiations. No of their supply chains.
every CEO’s priority list. technology or reassessment should be out of bounds
in the transformation from supply chain to network.
1 2 3 4 5 6 7 8 9 Supply unchained 61
1
can you deploy to capture and Explore the latest insights on
analyze more internal and external disruptive changes in supply chain,
data from across the supply and and learn how your organization
value chains? can respond.
2
How could you benefit from Learn how to drive efficiencies by
sharing information more freely scaling smart factory manufacturing
across your supply network? systems and responding to supply
chain stress.
3
Which nonrepetitive supply chain Delve into the research on how
tasks carry elevated safety risks? advanced manufacturing and IoT are
Which of these tasks could be connecting physical and digital.
performed by robots, cameras, or
other technologies?
1 2 3 4 5 6 7 8 9 Supply unchained 62
AUTHORS
Our insights can help you take advantage of emerging trends. If you’re looking for SENIOR CON T R IBU TOR S
fresh ideas to address your challenges, let’s talk.
Jennifer Brown Sarah Noble
Patrick Gallagher Hussain Mooraj Principal, Director,
Deloitte Consulting LLP Deloitte MCS Limited
Synchronized Planning & Supply Chain & Network Operations
Fulfillment leader chief futurist
Rafael Calderon Diogo Carneiro
Deloitte Consulting LLP Deloitte Consulting LLP
Principal, Senior manager, Deloitte &
patgallagher@deloitte.com hmooraj@deloitte.com
Deloitte Consulting LLP Associados, SROC S.A.
ENDNOTES
1. Brent Schrotenboer, “Coronavirus and shopping 9. Ibid.
for supplies: Getting to the bottom of the toilet 10. Deloitte, Using autonomous robots to drive supply
paper shortage,” USA Today, April 8, 2020. chain automation, 2017.
2. Procurious, “How now? Supply Chain Confidence 11. Adapted from an original case study created by
Index,” accessed November 5, 2020. Deloitte Development LLC, 2019.
3. Bernard Avishai, “The pandemic isn’t a black swan 12. Eric Piscini et al., Blockchain to blockchains: Broad
but a portent of a more fragile global system,” New adoption and integration enter the realm of the
Yorker, April 21, 2020. possible, Deloitte Insights, December 5, 2017.
4. Procurious, “How now? Supply Chain Confidence 13. Anthony Jackson and Michael Qualis, “Strategies
Index.” to counter supplier risk, preserve cash,” Deloitte
5. Jean-Paul Rodrigue, The Geography of Transport CFO Journal, September 8, 2020.
Systems, fifth edition (New York: Routledge, 2020).
MLOps: Industrialized AI
STRENGTH IN NUMBERS
TREND 4
MLOps: Industrialized AI
Scaling model development and operations with a dose of engineering and operational discipline
S
ophisticated machine learning models clunky, brittle development and deployment the primary reasons for failure.3 Many more
help organizations efficiently discover processes that stifle experimentation and projects (47%) fail to even make it out of the
patterns, reveal anomalies, make hinder collaboration between product teams, experimental phase and into production.4
predictions and decisions, and generate operational staff, and data scientists. In one
insights; Forrester reports that more than survey of nearly 750 business decision-makers, Many organizations are constrained by
half of global data and analytics technology only 8% considered their companies’ ML artisanal development and deployment
decision-makers have implemented or are in programs to be sophisticated. And deployment techniques, with star data scientists frequently
the process of implementing some form of happens too slowly: Twenty-two percent said it treated as virtuosos and given considerable
AI.1 As machine learning and AI increasingly takes between one and three months to deploy creative control. Typically, these models
become key drivers of organizational a newly developed ML model into production— are developed and deployed using manual,
performance, enterprises are realizing where it can deliver business value—with customized processes that, however clever,
the need to shift from personal heroics to another 18% saying that it takes more than aren’t terribly scalable. And enterprise data
engineered performance to more efficiently three months.2 infrastructure is not designed to support
move ML models from development through to rapid, consistent, streamlined development
production and management. As a result, IDC reports, 28% of AI/machine of machine learning models, as the chapter
learning projects fail, with lack of necessary Machine data revolution discusses.
Despite growing ML adoption, many expertise, production-ready data, and
organizations are hamstrung in their efforts by integrated development environments cited as
1 2 3 4 5 6 7 8 9 MLOps: Industrialized AI 66
Organizations may need to rethink cultural era of artisanal AI must give way to one of Today, it’s AI’s turn for the DevOps treatment.
norms, organizational structures, and automated, industrialized insights. Enter MLOps is an approach that marries and
governance mechanisms to more efficiently MLOps, also known as ML CI/CD, ModelOps, automates ML model development and
leverage AI resources, according to Jeff Butler, and ML DevOps: the application of DevOps operations, aiming to accelerate the entire
director of research databases at the Internal approaches and tools to model development model life cycle process. MLOps helps
Revenue Service. “AI and machine learning and delivery to industrialize and scale drive business value by fast-tracking the
can transform the way business is done, machine learning. experimentation process and development
but only if organizations can fundamentally pipeline, improving the quality of model
reshape organization structures, cultures, and production—and makes it easier to monitor
governance frameworks to support AI,” he MLOps optimizes development, and maintain production models and manage
improve collaboration among teams and Once models have been deployed to demand planning models weren’t updated
shorten development life cycles, thereby production and begin encountering more frequently enough to account for the quickly
enabling faster, more reliable, and more data, monitoring their performance can help emerging “new normal” as the pandemic
efficient model deployment, operations, and ensure they continue to deliver business value. began. As discussed in the chapter Supply
maintenance as well. If unchecked in production, unexpected bugs unchained, many businesses had either too
could be introduced into the pipeline. And as much or too little supply, in large part because
With automation and standardized processes, the data used to train and validate models their demand planning models were operating
MLOps can encourage experimentation ages, predictive accuracy can deteriorate. on data and assumptions that became
and rapid delivery, helping enterprises outdated nearly overnight.
Development focus shifts from But relying on a few experts has limits, chiefly
related to scalability and repeatability. Every
collaborative approach, enabling a broader
and larger team of professionals to work
exceptionalism to professionalism data Jedi typically prefers their own set of model together more efficiently to get more done in
development and deployment workflows, a standardized manner. Tools can help too:
Bringing the discipline of DevOps to machine based on education, experience, and personal Automated machine learning, or AutoML, can
learning can help AI adopters scale model preferences. They then often build models accelerate model development by helping
development and deployment, but they must with bespoke data extracts that can require data scientists quickly test different models
also tackle a significant skills gap. In a recent significant effort to recreate when later brought and variants.
Deloitte study, 68% of executives surveyed into a production setting using real-world, large-
described their organization’s skills gap as scale data. As machine learning permeates the These new players can help data scientists
“moderate-to-extreme,” with 27% rating it as enterprise, a more scalable, efficient, and faster test and fine-tune their creations, deploy
“major” or “extreme.”8 approach is needed to improve development models to production, manage production
resilience, reduce production bottlenecks, and models, address issues related to security and
Typically, enterprises rely on a small number increase the reach of ML projects. governance, and remove impediments to AI
of highly skilled data scientists and analysts and ML initiatives associated with outdated
to develop and test complex ML models and Organizations need supporting teams data infrastructures. Together with MLOps,
then deploy them to a production setting. of multitalented technology and ML data engineers and technologists can expand
With expertise in statistical analysis and professionals to help with activities such as the focus of AI teams from model building to
experience in determining appropriate ML data management, model deployment, and operationalizing. By lightening the load on the
approaches, developing models, making postdeployment monitoring and management. still-critical data scientists, the new supporting
prototypes, and ensuring the models’ MLOps practices encourage communication cast and crew can help ensure that the entire
predictive accuracy, these data scientists are between expanded development and production is as Oscar-worthy as the lead
in high demand. production teams; like DevOps, it’s a deeply actor’s performance.
1 2 3 4 5 6 7 8 9 MLOps: Industrialized AI 69
LESSONS
“To create truly transformational products, we need a more
The next wave of AI research consistent, synchronized approach to AI across the agency,”
which can serve as proofs of concept for benefit all line offices by efficiently leveraging Management Technology department. “There
larger-scale efforts. Another objective of the newest machine learning techniques when are limitations to doing everything manually as
NOAA’s AI strategy has been to strengthen scientifically appropriate.” long as data scientists and data analysts are
and expand partnerships in order to enhance working on their own ‘island’ without the ability
the use of AI to achieve the NOAA mission.11
Scaling to thousands of to collaborate or share data.”
In addition to partnerships and coordinating models in financial services Currently, the practice is using common
AI research, the NOAA AI Center is expected platforms for managing data and developing,
to be responsible for making NOAA’s data AI and machine learning technologies are deploying, and monitoring ML models. To
AI-ready and available to the agency and helping financial services firm Morgan build and test models, people created a
public, promoting ML algorithm development, Stanley use decades of data to supplement sandbox with access to a centralized data lake
AI labeling, application development, human insight with accurate models for that contains a copy of the data used in the
information exchange, and general AI fraud detection and prevention, sales and production system, a technique that makes it
awareness generation and workforce training. marketing automation, and personalized easier to bring models from development into
Technical specialists from the NOAA AI wealth management, among others. With an production.
Center, embedded in the line offices, will AI practice that’s poised to grow, the firm is
provide researchers with the know-how, leveraging MLOps principles to scale AI In the development environment, data
tools, and support to execute their ideas. and ML.12 scientists, business analysts, and data
“We’ll make sure to not stifle scientists’ engineers across the practice can access the
creativity and instead help them conserve “We need to be able to scale from hundreds of same standardized data in near-real time,
resources and enhance their use of AI when models to thousands,” says Shailesh Gavankar, enabling them to efficiently and collaboratively
needed,” Boukabara says. “By cross-fertilizing who heads the analytics and machine explore, prototype, build, test, and deliver
knowledge across the agency, we’ll be able to learning practice in Morgan Stanley’s Wealth ML models. Advanced techniques mask
1 2 3 4 5 6 7 8 9 MLOps: Industrialized AI 73
personally identifiable information so the As its AI practice evolves, Morgan Stanley to reimagine and reinvent critical back-end
teams can generate insights without exposing Wealth Management will be focusing on business processes. About two years ago,
sensitive data. “Across our AI practice, continuing to improve speed to market the company embarked on an AI-supported
processes are built around data accuracy and by further automating the model risk journey to streamline claims management. As
privacy,” Gavankar says. “Applying the highest management process and integrating the part of that process, leaders built a platform
standards to the training system ensures sandbox and production systems. “As that consolidates model development and
that we meet data compliance and regulatory MLOps tools and processes enable us to deployment across the enterprise.
requirements.” operationalize models more efficiently,”
Gavankar says, “we can continue to increase Anthem initially built several ML models
For good model governance, transparency, the number of models in production and that revealed patterns in claims data, made
and accountability, an independent, in- more fully leverage AI’s ability to drive better predictions to speed processing, and
house model risk management team was business decisions.” identified and corrected errors. The models
established. With years of experience were successful—and leaders realized they
The company built an integrated development deployed on the consumer side. “It’s much
environment and an end-to-end platform more efficient than when we used to develop
that serves as a one-stop shop where a model for each use case from scratch,”
developers and data scientists prepare and Arcot says.
store training data, build and validate models
via easy-to-use interfaces, and deploy them at Using the platform, Anthem data scientists
scale. A feedback mechanism allows models have developed a number of models, including
to continuously learn and improve while a those that fast-track the processing of pre-
separate platform monitors the performance approval claims, identify and automatically
of production models. reject duplicate claims, and determine
whether a medical procedure needs
Simultaneously, the company has been preauthorization. Previously, a human claims
working on an initiative that consolidates data examiner or clinician needed to manually
from seven systems into a single repository. review and process all of these claims.
With most of that work complete, the
process of finding the data to build, train, and Arcot says the platform has dramatically
operationalize models is much more efficient. increased model deployment speed. “Before
we developed the platform, it took about six
The platform also provides Anthem with the months to deploy very simple models,” he
flexibility to duplicate models for multiple use says. “Now we are able to develop much more
cases. For example, if a pipeline is already complex initiatives in half the time.”
built out into the legacy claims system for
a commercial use case, it can also be easily
1 2 3 4 5 6 7 8 9 MLOps: Industrialized AI 75
the lines of business, from engineering to be applied, they can meaningfully move the When it comes to ML adoption, data is often
product teams, people need a broader needle for the business. Without collaboration, cited as the No. 1 challenge. In our experience,
understanding of AI and ML concepts and AI teams risk building impressive prototypes more than half of the time spent building
tools to help identify relevant business that never get business buy-in or have real- ML models can involve data wrangling, data
opportunities and understand the potential world customer impact. cleanup, and pre-processing stages. If you don’t
of this technology for customers and other invest in establishing a strong data strategy,
key stakeholders. At Amazon, we addressed ML talent will be forced to spend a significant
this skills gap by building a Machine Learning Armed with an understanding portion of their time dealing with data cleanup
of AI fundamentals,
University in 2014.14 Available to anyone and management instead of inventing new
interested in machine learning, the university algorithms. Specifically, poor data and model
helps AI professionals keep their skills sharp
while giving product managers, program business stakeholders can governance are also significant challenges to
widespread AI adoption. Driven by concern
managers, and other novices the opportunity
to learn the basics of AI and ML.
play a collaborative role in that data will be used inappropriately, many
business divisions tend to hoard data into silos
provide feedback on how the model is changing companies invest in developing, deploying, enterprise, enabling organizations to efficiently
and alert human developers when the models and managing models. This undifferentiated ingrain machine learning into business
need to be retrained and recalibrated. heavy lifting can distract talent from value- processes and embed it into new products
driving tasks such as solving critical business and services.
In addition, a solid strategy for managing problems and building customer-focused
and storing data can help optimize data solutions. For many companies, a more
scientists’ skills and time. Automating time- efficient solution might be to leverage
consuming data management tasks can help existing platforms and tools, such as Amazon
free up these professionals to focus on what SageMaker, that expedite and simplify the
they do best: developing algorithms and model production process, drawing humans
building models. By simplifying the process into the loop for critical decision-making.
of classifying data and controlling access, Similarly, organizations nowadays do not
automated data management can help have to spend time building an automatic
address data governance challenges. speech recognition model for transcribing
contact center calls—instead, they can use
Finally, as businesses scale their ML cloud AI APIs, such as Amazon Transcribe, or
practices, it is important for builders to fully packaged AI products, such as Contact
focus on what is meaningful for the business Lens for Amazon Connect, that modernize
instead of worrying about developing ML contact centers.
infrastructure—an undifferentiated but heavy
workload. Streamlining model production can Removing these and other roadblocks
help organizations use their talent and other standing in the way of efficient ML adoption
resources more wisely. For example, many can help industrialize and scale AI across the
1 2 3 4 5 6 7 8 9 MLOps: Industrialized AI 78
EXECUTIVE PERSPECTIVES
STRATEGY // With ML adoption growing FINANCE // As organizations are RISK // ML deployments are quickly
across industries, CEOs—particularly those increasingly pressed to make good scaling up and enabling algorithms to
whose companies operate in low-growth decisions faster and develop better models make key decisions for the organization.
sectors—are exploring how to use machine learning for demand forecasting, finance leaders are quickly Yet trust remains an issue: Humans are undeniably
to grow market share and lower costs. CEOs may realizing that their organizations need machine prone to bias, but the press and the public often
want to speak to their CIOs and IT teams about their learning at scale. Assuming that technology speed take particular notice of biases in machines and
vision for applying AI/ML to boost the bottom line. and capability will continue to increase exponentially, biased outcomes of ML models.16 CROs can work
For example, if they hope to increase earnings per making a machine-based decision in the future will with their CIOs, CDOs, and other IT leaders to
share by 10 points, CEOs should make their priorities cost a fraction of a nonscalable human decision anticipate potential brand risks and suggest design
clear and spend time understanding what can be today. Indeed, 67% of executives in Deloitte’s State workarounds. They can also make purposeful
achieved and/or what investments are needed. As the of AI survey are already leveraging ML for efficiency choices with AI and ML algorithms not only to help
organization hires AI/ML talent to scale capabilities, gains, such as faster account reconciliation or more maintain public trust in their organizations but to
leaders should provide a clear mandate to these new accurate accruals.15 To ready their organizations for position risk management protocols for AI/ML as a
teams for how and when technology should augment this change, CFOs can choose between becoming competitive differentiator.
human decision-making. more technically savvy or buying financial planning
and analysis as a service. Whether they sponsor
data officers or take on the task themselves, finance
leaders may soon rely on the power of machine-
driven insights for their regular updates to analysts
and shareholders.
1 2 3 4 5 6 7 8 9 MLOps: Industrialized AI 79
1
Do you have the skill sets and See how companies are managing
organizational structure needed and getting ahead of the rise in
to meet your AI goals today? AI adoption.
In two years?
2
How can you improve the time to collection
market of models and improve Learn how cognitive technologies
their performance in production? can augment manual tasks and
decision-making.
3
governance, accountability, and to the enterprise
transparency? What precautions Listen to a podcast on the ways
can reduce developer and data MLOps can integrate AI/ML models
bias? How can you better protect into business processes.
sensitive data?
1 2 3 4 5 6 7 8 9 MLOps: Industrialized AI 80
AUTHORS
Our insights can help you take advantage of emerging trends. If you’re looking for SENIOR CON T R IBU TOR S
fresh ideas to address your challenges, let’s talk.
Paul Phillips
Beena Ammanath David Kuder Managing director,
Deloitte Consulting LLP
Executive director, Deloitte’s US AI Insights and Engagement
AI Institute Market offering leader
James Ray
Deloitte Consulting LLP Deloitte Consulting LLP
Senior manager,
bammanath@deloitte.com dkuder@deloitte.com
Deloitte Consulting LLP
ENDNOTES
1. Srividya Sridharan, Gene Leganza, and Jeremy the age of AI: Insights from Deloitte’s State of AI in 15. Deloitte, Why CFOs should have artificial
Vale, “Research overview: Artificial intelligence,” the Enterprise, 2nd Edition survey, Deloitte Insights, intelligence on their minds, 2020.
Forrester, May 20, 2020. March 3, 2020. 16. Craig S. Smith, “Dealing with bias in artificial
2. Algorithmia, 2020 state of enterprise machine 9. Sid-Ahmed Boukabara (principal scientist for intelligence,” New York Times, November 19, 2019.
learning, 2019. strategic initiatives, NOAA’s Center for Satellite
Applications and Research), phone interview with
3. IDC, “IDC survey finds artificial intelligence
authors, October 14, 2020.
adoption being driven by improved customer
experience, greater employee efficiency, and 10. National Oceanic and Atmospheric Administration,
accelerated innovation,” June 10, 2020. NOAA artificial intelligence strategy, February 2020.
4. Jessica Davis, “Getting machine learning into 11. National Oceanic and Atmospheric Administration,
production: MLOps,” Information Week, June 26, “AI agreement to enhance environmental
2019. monitoring, weather prediction,” October 27,
2020.
5. Jeff Butler (director of research databases at the
Internal Revenue Service), phone interview with 12. Shailesh Gavankar (head of Analytics and Machine
authors, October 16, 2020. Learning practice, Morgan Stanley), phone
interview with authors, October 16, 2020.
6. Nitin Mittal, Dave Kuder, and Samir Hans, AI-fueled
organizations: Reaching AI’s full potential in the 13. Harsha Arcot (senior director of enterprise data
enterprise, Deloitte Insights, January 16, 2019. science, Anthem), phone interview with authors,
October 23, 2020.
7. Cognilytica, “The rapid growth of MLOps,” April 2,
2020. 14. Amazon, “Machine Learning,” accessed November
23, 2020.
8. Susanne Hupfer, Talent and workforce effects in
82
AMPLIFY ML CAPABILITIES
TREND 5
W
ith machine learning (ML) poised to including advanced data capture and capabilities, are already monetizing more
augment and in some cases replace structuring capabilities, analytics to identify diverse data, more quickly.2 Importantly, end
human decision-making, chief connections among random data, and next- users have less and less patience for the
data officers, data scientists, and CIOs are generation cloud-based data stores to support kind of latency that legacy systems and data
recognizing that traditional ways of organizing complex modeling. models often deliver. The optimum latency
data for human consumption will not suffice time between click and desired response is
in the coming age of artificial intelligence (AI)– Some companies are already embracing this less than 50 milliseconds—any longer and
based decision-making. This leaves a growing trend as part of larger AI initiatives. In Deloitte’s users become irritated by the delay and make
number of future-focused companies with third annual State of AI in the Enterprise survey, “executive decisions” themselves.3
only one path forward: For their ML strategies to when asked to select the top initiative for
succeed, they will need to fundamentally disrupt the increasing their competitive advantage from
data management value chain from end to end. AI, respondents singled out “modernizing our Humans, machines, and data
data infrastructure for AI.”1
In the next 18 to 24 months, we expect to see In the coming months, participants in the
companies begin addressing this challenge For digital nonnatives participating in this machine data revolution trend will explore
by reengineering the way they capture, store, trend, the stakes are high. Some of their opportunities to reengineer their data
and process data. As part of this effort, they digital native competitors, largely unburdened management value chains to support ML’s
will deploy an array of tools and approaches by outmoded data models and processing possibilities. In the arena of data management,
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 84
this marks a distinct change of course. For recommendations—ML may only need to Though approaches can vary by industry,
decades, companies have collected, organized, make good enough decisions, not perfect ones. market, and organizational need, many
and analyzed data with one goal in mind: For example, a retailer would presumably trend participants will likely focus on their
helping humans make decisions based on see value in the ability to recommend, in real reengineering efforts on the following areas:
statistical fact rather than hunches and time, an assortment of products tailored very
emotion. Humans tend to look at aggregated broadly to thousands of individual online
data characterized by two or three major shoppers simultaneously. The products that Capture and store
factors. When faced with more complex ML algorithms recommend might not perfectly
data, many humans struggle to process the match each customer’s unique tastes, but Chances are, your organization has troves of
information presented and to articulate a they might be sufficient, in that moment, to data that’s potentially valuable yet untapped.
useful decision. As such, we typically organize drive a sale. Across an enterprise, each good- Some of it is probably traditional enterprise data
data for humans in clean tables and rows, enough data-based decision that machines residing in databases, files, and systems; other
with precise labeling. Machines, by contrast, make, rather than humans, drives down the troves may be more recent data generated by
can assess multiple factors simultaneously overall cost per decision, which in turn enables machines or mobile devices. Still others may be
and objectively. ML models can extract low companies to extract value from even the unstructured text, or nontraditional data from
levels of statistical significance across massive lowest-level decisions. Creating an automated video or audio recordings. In all likelihood, this
volumes of structured and unstructured data. pipeline that replaces low-level or nonscalable data was previously too hard or too expensive
They work around the clock and can make human decisions with those made by to capture and utilize in a cost-effective way,
clever decisions in real time. machines brings to mind the promise of so it lies fallow. This is a lost opportunity. No
Moore’s Law. Over time, speed and capability one knows which data amid vast stores of raw
When used in areas in which human decision- will increase so dramatically that making that information might turn out to be predictive or
making is nonscalable—such as cleaning up data-based decision in the future will cost a confer some decisioning value down the line,
raw data4 or making personalized product fraction of what it does today. so it is critical to capture all the data you can.
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 85
Moreover, you are probably throwing out from IoT, social media, and AI in a variety of forecasts the data warehouse-as-a-service
some data today that, with the right tools modern database technologies, including: market will reach US$23.8 billion in value
and approaches, you can use. Take utility by 2030.6
companies, for example. What information • Cloud data warehouses. The cloud-
do they need to predict power or equipment based data warehouse, which a growing • Feature stores. In the near future, it
failures? Traditionally, they may have collected array of major and emerging public will be commonplace for an organization
data only on failure. But for predictive cloud vendors are offering as a service, to have hundreds or thousands of data
purposes, they would also need data on aggregates data from disparate sources models operating independently of
uneventful everyday operations to understand across an enterprise and makes it each other, and in parallel. Each of these
what normal looks like. This same idea applies available to users for real-time processing models will use different feature sets.
to people visiting your company’s website. Do and mining. This permissions-based, For example, some require immediate
you have website data for both success and centralized system eliminates the need decisions while others do not, thus placing
failure? In a world where data quality no longer for colocated data and data pipelines. broadly different demands on data and
matters as much as it once did, what changes In addition to collation and storage on processing power. Pushing real-time
can you make to your current data practices capabilities, cloud data warehouses also compute uniformly in every model is a
to make them more predictive? typically offer search engine tools for waste of computing power. Likewise, some
querying data and analytics capabilities.5 models probably share features while
In terms of storage, organizations are This combination of public cloud ease- other features may be used exclusively
becoming less focused on storing clean data of-use, the ability to scale up or down as in a single model. How can you manage
that fits neatly into tables, rows, and columns. needed, and advanced data processing all of these competing demands across
To feed ML algorithms and advanced analytics and analysis tools is fueling considerable data models? Feature stores provide a
tools, many are exploring opportunities to growth in the cloud data warehouse mechanism for allocating compute, sharing
store massive volumes of unstructured data market. Prescient & Strategic Intelligence features, and managing data efficiently,
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 86
and at scale, which makes this mechanism • Graph databases. Highly interconnected more, modern self-healing, fault-tolerant
integral to driving down decision costs. data can be challenging to analyze data architecture typically requires less
What’s more, by leveraging AI, feature and use to its fullest potential. Using maintenance, which can reduce administrative
stores may eventually be able to predict traditional relational databases in which and repair costs. Thus, the potential benefit of
demand of certain features based on the data are organized in tables, one can increasing storage capacity could far outweigh
types of data being modeled.7 identify and manage a limited number whatever costs you may incur. ML and
of data relationships. But as data grows advanced analytics can discern low levels of
• Time series databases. The popularity more voluminous and less structured, statistical significance across a large number
of time series database technologies has the number of relationships and of factors, which in turn can provide a
grown considerably over the last two interconnections increases exponentially, significant lift that would be near impossible
years, with good reason.8 Unlike relational thus becoming unmanageable (and to achieve using traditional data storage and
databases that record each change to data unsearchable) in traditional database modeling techniques.
as an update, time series databases track models. Graph databases are designed
and record them—and the specific time
they were made—as a unique insert into
specifically to address this challenge by
storing not only data but information Organizations are
a dataset. With the explosion of temporal
data from IoT and monitoring technologies,
on each data point’s relationships in a
native way. With this model, queries about
becoming less focused
among others, both historical and complex relationships among data can be on storing clean data
predictive analysis increasingly depends
on the ability to query a data value from
fast, efficient, and more accurate.10
that fits neatly into
one point in time and track it continuously,
accurately, and efficiently.9
With storage costs continuing to fall,
aggregating and organizing massive volumes
tables, rows, and
of data is no longer cost-prohibitive.11 What’s columns.
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 87
Discover and connect if you can collect all known and unknown • Analytics, semantic models, and cognitive
enterprise data, how can you tie these disparate, technology can automate manual, costly
As you begin capturing more data, it will likely inconsistently formatted and named data points stewardship activities—thus freeing up
include fragmented data generated across together in a way that is meaningful? The work data scientists to focus on more advanced
different devices, channels, and geographies. of discovering and connecting enterprise data analysis.
How can you connect fragmented data can be formidable and costly. Yet shirking this
in a way that characterizes an individual challenge could cost even more if your company • Identifying similarities in underlying data
customer in an individual context—or reveals misses out on potentially valuable opportunities. systems’ code makes it possible for data
an unmet need the marketplace or an scientists to use custom data algorithms in
internal opportunity for greater efficiency?
Unlocking the full value of all data resources,
To make the strongest multiple data models.
including dark and nontraditional data, can data-driven decisions, • Finally, by leveraging ML capabilities to
data steward deployments to expand further compute and processing power away from of devices they can effectively manage. 5G can
into transactional systems, supply chain a centralized point and closer to a network’s deliver faster speeds and millisecond latency.
ecosystems, and smart factory environments. “edge” or periphery. It does not replace It can also expand bandwidth capacity to
enterprise or cloud-based data centers but simultaneously manage many more devices
helps distribute processing work—including per square kilometer.14
Serving up ML’s secret sauce analysis and decisioning—more evenly across
a network. Rather than sending raw data back
The ability to process larger volumes of diverse to a cloud or data center, a device operating The way forward
data in real time is the secret sauce of ML- at the edge generates action independently
based data decisioning. The faster that big data or sends only already-refined data to the As we discuss in this report’s MLOps:
systems can capture and process data, feed it network, in effect storing, processing, Industrialized AI chapter, ML initiatives are
into ML and analytics platforms, and then serve analyzing, and reacting locally. Edge computing gaining momentum across industries.
up insights to users, the more impactful your can be particularly useful when deploying ML Indeed, the ML technologies market is
data investments can be. algorithms, which require uninterrupted, real- currently growing at a rate of 44% annually
time access to large quantities of recent data.13 and is expected to reach US$8.8 billion
To this end, a growing number of organizations in value by 2022.15 But ML algorithms and
are exploring ways to make decisions at data’s Advanced connectivity also has an important platforms will deliver little ROI in companies
point of entry into the network rather than enabling role to play in real-time decision- with outdated data infrastructure and
sending it first to the core or cloud. Some making at data’s first point of entry. Current processes that were leading-edge in 2002.
are building edge computing capabilities that generation connectivity, such as 4G/LTE and
can decrease latency in data systems while Wi-Fi, can support some edge computing and How will you reengineer your data
also making these systems more reliable and real-time data processing needs, but they are strategies to build a new foundation for
efficient. Edge computing means pushing limited by bandwidth, latency, and the number your company’s future?
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 89
LESSONS
from prototype to full scale production for ML models. These
Adventures in data democracy models have had dramatic results, such as blocking 6.5 billion
FROM THE Since its founding 144 years ago, AT&T has reinvented itself
robocalls to customers, deterring fraud in AT&T stores, and
making technicians visits to customer homes more efficient.
FRONT LINES many times to harness historic disruptive innovations such as
the transistor, the communication satellite, and more recently, AT&T started its data transformation journey in 2013 when it
the solar cell.16 Today, the global technology, media, and began aggregating large volumes of customer and operational
telecommunications giant is reinventing itself again—this time data in data lakes. In 2017, the company created a chief data
as a pioneer in the use of ML, which it is deploying broadly in office with the goal of leveraging these rapidly growing data
areas such as IoT, entertainment, and customer care.17 stores for “hyper-automation, artificial intelligence, and machine
learning.” The ongoing work of achieving these goals has
The company is also leveraging ML to reimagine the way presented several significant challenges. First, in a company
it finds, organizes, and uses data. “One of the things we as large as AT&T, it was sometimes difficult to find and access
wanted to do was automate some of the routine cleansing potentially valuable data residing in legacy systems and
and aggregation tasks that data scientists have to perform databases. And even when data scientists eventually found
so they could focus on more sophisticated work,” says Kate such data, they occasionally struggled to understand it, since
Hopkins, vice president of data platforms, AT&T Chief Data it was often labeled inconsistently and offered no discernable
Office.18 Likewise, the company wanted to develop a way context or meaning. Finally, there was a formidable latency
to democratize meaningful data, to the extent consistent challenge across all data systems that, left unaddressed, would
without privacy, security, and other data use policies, making stymie the real-time data needs of ML models.
it more broadly available to qualified personnel across the
enterprise. These efforts, Hopkins says, have already borne To address these challenges, the chief data office developed
fruit. New tools have shrunk the time to market required to go the Amp platform. Amp enables a culture of technology
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 90
and data-sharing, reusability, and extensibility data and tools to the public cloud. Working
at AT&T. Pari Pandya, director of technology closely with cyber teams to ensure data and Data and IT double-team
and project manager for Amp, says that what IP security, the company is leveraging the digital transformation
began a few years ago as an internal online cloud’s ability to scale up compute power as
marketplace (aggregating microservices, needed. The cloud’s power is helping create How can a 100-year-old retail organization
APIs, chat bots, designs, etc.) for accelerating the real-time access that ML—as well as efficiently and accurately take data from legacy
automation, has evolved into a single, powerful enterprise stakeholders and customers— applications that were designed for very
source of data truth for systems and users. require. Unlimited access to compute on specific use cases to accomplish something
Consider this: As data flows through multiple demand through the cloud and the availability that those applications were never intended
systems and processes, its definitions change. of business-ready data is accelerating the to do?
Amp not only finds legacy system data, it uses journey.
metadata to ascribe meaning to this data, and “Every legacy company faces this challenge,”
provides a clear lineage to help users better Hopkins notes that AT&T’s data transformation says Paul Ballew, chief data and analytics
understand the data. “It serves as a business journey has yielded another welcome benefit. officer at the leading Canadian food and
intelligence platform that provides not only “The business units have become much more pharmacy retailer Loblaw.20 “You have to bring
meaningful data but analytic and visualization knowledgeable about data science and are those data assets together from across your
tools that empower business teams, strategists, identifying opportunities to use data in new ecosystem in a way that’s scalable, repeatable,
and product developers to leverage data in ways. Across the board they’re requesting and governable, which is no small task.”
more advanced ways and share insights through much more mature and sophisticated
data communities,” Pandya says.19 data,” she says, adding that “being able to Taking an ecosystem approach to data is
democratize data and make the process particularly formidable in a successful retail
To meet the challenge of latency, AT&T is transparent across the enterprise can deliver organization such as Loblaw, which operates
on a multiyear journey to move some of its exponential payback.” 2,400 stores and maintains an expansive
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 91
e-commerce presence. “We are a legacy From a technology standpoint, Loblaw takes a “Once you’ve coordinated these three layers,
company trying to leverage technologies that three-layered approach: you have to manage the analytic solutions,
digital natives are born with,” Ballew says. refresh cycles, and monitor to ensure strong
• Data layer. An array of data management and consistent governance,” Ballew says,
Yet despite its challenges, data represents a and digital capabilities that make data noting that because Loblaw does a variety
unique opportunity on the path to Loblaw’s assets, many from legacy applications, of things from selling groceries to providing
digital future. And unique opportunities consumable in near-real time to a variety pharmacy and health services, the company
require unique approaches. Like many digital of complex use cases. has a sliding scale in terms of data sensitivity,
nonnatives, the company is shifting its focus privacy concerns, and legal compliance. As
from traditional data management priorities • Analytics and development layer. A such, robust data governance is critical in
such as storage, curation, and quality to a collection of AI/ML and advanced analytics protecting sensitive data and determining
new, more complex arena in which data technologies that bring data assets to model sensitivity to bias and other factors.
analytics and digital solutions drive day-to-day life and glean insights from structured “We have to be proactive stewards of
operations. “It requires a different approach to and unstructured data to support better customer data and leverage it in a manner
‘baking the soufflé,’” Ballew says. “We source decisions and more efficient workflows. that results in providing benefits to them in a
and mix ingredients differently, and then serve transparent manner,” he says.
it in new ways to those consuming it.” • Solution delivery layer. Provides tools
to systematically integrate decisions and Data’s ascent as a decision-driving, business-
Recognizing the critical importance of data insights into processes and applications, critical asset has redefined the roles of
in the company’s digital future, Loblaw set helping meet the organization’s digital Loblaw’s data and IT teams. “Our work around
up a distinct data organization that works in strategy goals. data and digital helps the business make
tandem with IT to drive digital transformation critically important decisions: who to talk to,
and engage the business. how to optimize marketing, run a factory,
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 92
or engage customers—the list goes on,” that enables data scientists to request data Pillai says. “Particularly in the post-COVID era,
Ballew says. “In terms of data, we helped the quality issues be fixed at the source and focus companies can’t meet face-to-face with clients,
organization and leadership understand the on turning data into value. “In the past, data so they rely more heavily on data and analytic
art of the possible and the implications—good scientists would find a problem, fix it, and keep insights. The analytics capabilities we have in
and not-so-good—of taking a comprehensive going,” says Santhosh Pillai, chief architect and place deliver these insights and unleash the
approach. The change has been beneficial data management. “Now they can provide value contained within our data.”
overall, but it has impacted our entire feedback to the source where data is mined,
ecosystem and those working in it.” and say, ‘do it differently.’ Over time, data quality The third component of ABN AMRO’s data
improves, and data scientists don’t have to transformation effort is a multifaceted data
And Ballew’s advice to those managing similar spend as much time on cleansing mesh model that moves data anywhere
change in their own companies? “Seek to and querying.”21 it needs to go within the ecosystem, from
understand before you seek to be understood.” source all the way to consumer. This “data
Strengthening governance at the source is just supply chain” serves not only as a distribution
to be event-driven—or for current data replicating data that is stored in several clouds
usage patterns. Today, algorithms and end and even moving that data between multiple
users read up-to-the-minute data far more cloud vendors.”
frequently than they use it in transactions.
Legacy data management models were not
designed to respond to constant read queries
and real-time updates.
Consider, for example, our trucks. Each Daimler truck that rolls off
the factory floor is a new digital asset. An array of onboard sensors
and other technologies monitor vehicle performance continuously,
generating data that offers Daimler real-time insights on a truck’s
health. But rather than simply providing the vehicle owner with a
status report, we can now apply analytics to vehicle performance
data to predict when a part might fail. In urgent cases, we share this
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 95
information with the vehicle owner and initiate opportunities. In the intelligent-company model, than ever. In addition to managing the data
a “service now” event, directing the owner to our relationships with customers deepen after itself, there are more and more regulations
a nearby service facility where mechanics— the vehicle purchase. We use technology and governing data use, and we must understand
made aware of the problem and confirmed to data to refine existing products and services which data can be used and which services
have the needed part in stock—can address to meet unique needs as well as to create new can be provided while always maintaining
the issue without delay. services around the vehicle. This is a completely regulatory compliance. Likewise, we must
different way of working—one that we are clearly understand customer preferences
The ability to provide this kind of predictive embracing wholeheartedly. and expectations for the way we use their
vehicle performance data presents an data, which also affects the way we offer
opportunity to create an entirely new value- services. Should we offer individual services,
added service. Now, because we can predict Data presents an opportunity or can we bundle them? How do different
EXECUTIVE PERSPECTIVES
STRATEGY // CEOs are rarely involved at FINANCE // As the capabilities of RISK // In the Age of With,22 data
the deeper levels of data management. modeling and ML become more robust strategies focus on providing analysis and
But those whose companies are deploying (as detailed in our MLOps trend), finance insights that augment human decision-
ML and other data-dependent AI tools may want leaders are leaning into data stewardship and even making. In this construct, algorithms—rather than
to spend some time thinking about data strategy. sponsoring data governance roles. With better data data scientists and executives—may be consuming
Existing enterprise systems and data pipelines were comes better modeling, which means CFOs can more raw and unstructured data, but the inherent risks
not designed with AI needs in mind, so CEOs can accurately produce internal reports, predict quarterly associated with managing large volumes of data
engage their IT teams to lay the groundwork for this performance, and shape the perceptions of market remain constant. CROs can play a key role in defining
shift. They should hold the CIO or chief data officer analysts and shareholders. However, the data needed risk parameters around protection, confidentiality,
accountable to carry out preparations for turning the for good modeling is not always internally owned— and integrity of large volumes of ML-optimized data.
organization’s MLOps vision into a reality. They can consider the example of parts manufacturers They can also develop separate risk strategies for
also expect ongoing investments, as data is gradually that monitor flight statistics to predict aluminum more sensitive customer data and determine whether
prepared for algorithms first instead of humans. demand. Such combinations of raw, external data and how such data can be used in ongoing MLOps
These new investments in data can eventually be with company data can lead to actionable insights efforts. By delving deeper into these issues, CROs
paid back through increased profits or lowered cost amid uncertainty. To enable these insights in the next can help their organizations become more deliberate
per decision. 18–24 months, CFOs may need to think outside of the about where to apply AI in the business and, perhaps,
box to source new types of data, enable agile model explore opportunities to monetize risk data with new
governance, and optimize for modeling outcomes. products and services.
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 98
1
How have you reevaluated data AI success
governance and data management Learn about the challenges to
as you move to increase the use of preparing data for AI initiatives and
AI and ML? where businesses can find support.
2
What mechanisms support artificial intelligence
accessing data from key legacy Read about the many ways AI
systems? Do they meet your can augment standard business
current and future needs? processes for better outcomes.
3
Which untapped sources of data taught us about data strategy
could have the greatest impact on Study the key lessons learned from
your decision-making? governments that have used data to
combat the coronavirus pandemic.
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 99
AUTHORS
Our insights can help you take advantage of emerging trends. If you’re looking for SENIOR CON T R IBU TOR S
fresh ideas to address your challenges, let’s talk.
Edward Bowen Matt Iames
Paul Phillips Sandeep Sharma, PhD Managing director, Specialist master,
Deloitte & Touche LLP Deloitte Consulting LLP
Hux chief technology officer Deputy chief technology officer
Deloitte Consulting LLP Deloitte Consulting LLP
Carl Gerber
paulphillips@deloitte.com sandeepksharma@deloitte.com Managing director,
Deloitte & Touche LLP
Irfan Saif Juan Tello
US AI Strategic Growth offering leader Chief data officer Adarsh Gosu
Deloitte & Touche LLP Deloitte Consulting LLP Managing director,
Deloitte Consulting LLP
isaif@deloitte.com jtello@deloitte.com
Jason Price
Managing director,
Deloitte & Touche LLP
Piyush Sacheti
Senior manager,
Deloitte Consulting LLP
1 2 3 4 5 6 7 8 9 Machine data revolution: Feeding the machine 100
ENDNOTES
1. Beena Ammanath, Susanne Hupfer, and David are they critical for scaling data science?,” Towards 15. MarketsandMarkets, Machine learning market by
Jarvis, Thriving in the era of pervasive AI: Deloitte’s Data Science, April 7, 2020. vertical (BFSI, healthcare and life sciences, retail,
State of AI in the Enterprise, 3rd Edition, Deloitte telecommunication, government and defense,
8. DB-Engines, “DBMS popularity broken down by
Insights, July 30, 2020. manufacturing, energy and utilities), deployment
database model,” November 2020.
mode, service, organization size, and region—global
2. Suketu Gandhi et al., “Demystifying data 9. Matt Asay, “Why time series databases are forecast to 2022, September 2017.
monetization,” MIT Sloan Management Review, exploding in popularity,” TechRepublic, June 26, 2019.
November 27, 2018. 16. AT&T, “Corporate profile,” accessed November 9, 2020.
10. Serdar Yegulalp, “What is a graph database? A
3. Samuel Newman, “What is latency and how 17. AT&T, “AT&T’s machine learning platform,” YouTube,
better way to store connected data,” InfoWorld,
does it kill fast internet?,” BroadbandDeals.co.uk, February 1, 2017.
March 21, 2018.
March 30, 2017. A decade ago, Amazon estimated
18. Kate Hopkins (vice president of data platforms, AT&T),
that every 100ms of latency costs the company 11. Robin Harris, “Data storage: Everything you need to
phone interview with authors, September 8, 2020.
1% in sales. See: Yoav Einav, “Amazon found know about emerging technologies,” ZDNet, May
every 100ms of latency cost them 1% in sales,” 28, 2019. 19. Pari Pandya (director of technology and Amp project
manager, AT&T), phone interview with authors,
GigaSpaces, January 20, 2019. 12. Deloitte CIO Journal on the Wall Street Journal, “Tips for
September 8, 2020.
transforming data management.”
4. Deloitte CIO Journal on the Wall Street Journal, “Tips
20. Paul Ballew (chief data and analytics officer, Loblaw),
for transforming data management,” August 24, 13. Edmond Toutoungi, “Cloud and edge computing
phone interview with authors, October 6, 2020.
2016. explained … in under 100 words,” Deloitte, August
27, 2019. 21. Santhosh Pillai (chief architect and data
5. Sean Michael Kerner, “Top 8 cloud data
management, ABN AMRO), phone interview with
warehouses,” Datamation, September 10, 2019. 14. Chris Arkenberg et al., Unbundling the cloud with
the intelligent edge: How edge computing, AI, and authors, October 16, 2020.
6. Prescient & Strategic Intelligence, “Data
advanced connectivity are enabling enterprises to 22. Deloitte, The Age of With: Exploring the future of
warehouse as a service market,” March 16, 2020.
become more responsive to a fast-moving world, artificial intelligence, November 7, 2019.
7. Adi Hirschtein, “What are feature stores and why Deloitte Insights, September 8, 2020.
101
TREND 6
C
onventional castle-and-moat With cloud now mainstream, businesses Consider the case of an employee who logs
cybersecurity models, which rely on managing services across multiple cloud in regularly on weekdays from her home
secure network perimeters and virtual providers are responsible for securing these and occasionally on weekends from a coffee
private network-based employee and third- technologies. As an enterprise more frequently shop. When her username and password are
party remote access, are proving to be no relies on third-party vendors to host and used on a Saturday night from somewhere
match for evolving cyberthreats, particularly manage data, infrastructure, and other services, in Eastern Europe, traditional approaches
as business models and workforce dynamics the attack surface expands. In one study, 59% might allow the connection. But because a
evolve. For instance, the move to cloud of companies surveyed had experienced a zero trust architecture is more risk-driven and
and hybrid IT environments—along with data breach due to a vendor or other third context-aware, it recognizes the inconsistency,
increasing numbers of cloud-based systems, party;1 another study concluded that multiparty automatically denies the access request,
remote workers, and connected devices— security incidents result in 13 times the financial and raises an alert. Automated response
are constantly expanding and dissolving the losses of single-party events.2 capabilities could be triggered to temporarily
network perimeter. The anticipated growth disable the user’s account, given the likelihood
of smart devices, 5G, edge computing, and In fact, while perimeter-based security assumes that its credentials have been compromised.
artificial intelligence promises to create even the trustworthiness of users and devices
more data, connected nodes, and expanded connected to the organization’s network, stolen Proper design and engineering of zero trust
attack surfaces. credentials cause more than a quarter of architectures can result in simple, modular
security breaches.3 environments and straightforward user access
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 103
control and management. Streamlining the applications, workloads, and other resources By removing the assumption of trust from
security stack can eliminate considerable into individual, manageable units to contain the security architecture and authenticating
management headaches, significantly reduce breaches and wrap security controls at the every action, user, and device, zero trust helps
operational overhead, and help scale to tens lowest level possible. By limiting access based enable a more robust and resilient security
of thousands of users. Similarly, onboarding on the principle of least privilege, a minimum posture. The organizational benefits are
employees, contractors, cloud service set of users, applications, and devices has complemented by a considerable end-user
providers, and other vendors can become access to data and applications. perk: seamless access to the tools and data
more efficient, flexible, responsive, and secure. needed to work efficiently.
Carefully designed zero trust architectures The anticipated growth As the benefits of zero trust continue to pile
assets; companies that lag on basic cyber security issues across the complete attack They can extend their operations beyond
hygiene principles and practices may be surface. To facilitate risk-based policy traditional boundaries while protecting
challenged to realize the full benefits of zero decisions surrounding their assets, it’s critical resources and maintaining an
trust. Fundamentals include: critical for organizations to understand the efficient user experience by moving the
enterprise IT environment. identity stack to the cloud, consuming
• Data discovery and classification. identity-as-a-service, or implementing
Data governance, inventory, classification, • Configuration and patch management. such advanced authentication methods as
and tagging are critical. To create the Without the ability to efficiently manage and physical biometrics, behavioral monitoring,
appropriate trust zones and access document baseline configurations of key and conditional access.
controls, organizations need to understand technology systems, deploy appropriate
their data, the criticality of that data, where patches, test patched systems, and • Third-party risk management. To
it resides, how it is classified and tagged, document new configurations, companies fully understand their entire risk surface,
and the people and applications that cannot easily identify changes and control organizations need greater visibility
should have access to it. risks to these systems. Malicious actors can into cyber risks related to their supply
exploit any vulnerabilities to gain a foothold chains and ecosystem partners, including
• Asset discovery and attack-surface within an organization. suppliers to third-party vendors.
management. Many organizations lack
a real-time, updated inventory of all IT • Identity and access management. • Logging and monitoring. To identify
resources—including cloud resources, To ensure that access to technology potentially malicious incidents and issues,
IP addresses, subdomains, application resources is granted to the proper people, security teams need automated logging
mapping, code repositories, social media devices, and other assets, enterprises and monitoring systems with advanced AI
accounts, and other external or internet- need to standardize and automate their and machine learning capabilities to help
facing assets—and therefore can’t identify identity life cycle management processes. simplify the process of tracking, analyzing,
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 105
and correlating data from volumes of requirements, and behaviors—warrants a vulnerabilities, enabling staff to keep pace with
detailed logs as well as alerts generated more automated approach to monitoring, incoming alerts and notifications, improving
by internal and external systems, security decision-making, enforcement, and auditing. operational efficiency and accuracy, and
controls, networks, and processes. decreasing response time.
Many existing security technologies can be
and orchestration
orchestration, zero trust adopters should
rationalize the security stack and eliminate
technologies can be
Many security operations center (SOC) teams
unnecessary and duplicative technologies
or those that contribute to data overload,
leveraged to build out
are challenged to keep pace with the volume of delay detection and response, and complicate zero trust architectures.
information generated by their technology and system maintenance and management.
security controls. They must monitor, manage, “Migrating to zero trust architectures can seem
and act upon continuous alerts and streams With a simplified security stack, existing like a heavy lift, especially in large enterprises
of data generated by fragmented security systems and tools can be integrated via API saddled with legacy technologies and a lot
architectures and disparate, disconnected tools. connections to a security orchestration, of technical debt,” says a senior technology
automation, and response (SOAR) platform leader at a large financial institution. “You
The number and nature of risk factors that can automate workflows and repetitive have to break it into manageable chunks
interrogated to support zero trust and manual tasks, and coordinate the flow of where you can identify a discrete win, such as
authentication and authorization—users, data and alerts to the SOC. SOAR platforms deploying pervasive endpoint segmentation,
devices, or credentials and contextual data help add context to triggered events and and understand that win as part of your larger
points such as location, privileges, application can auto-remediate identified and known story of operationalizing zero trust.”
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 106
LESSONS
Previously, access to an internal application would require
Zero trust mindset enables granting access to the Takeda network, which inherently
FROM THE digital growth enabled access to a number of additional, unrelated services.
“We could have tried to manually manage and restrict this
FRONT LINES Like many global pharmaceutical companies, Takeda
Pharmaceuticals supports better patient outcomes by
additional system access, but, invariably, things will be missed
over time,” says Scott Sheahen, global head of information
innovating and collaborating across a diverse group of internal risk management. “With the zero trust approach, we eliminate
and external stakeholders—including, in its case, more than superfluous system access and thereby reduce the avenues
52,000 employees and thousands of research partners, that could be exploited in a future cyberattack. Now we have
logistics partners, and other third-party service providers as granular, policy-based controls so that people have access
well as patients, physicians, and other health care providers. only to needed resources.”
The ongoing need to extend access to applications and
systems to its broad external ecosystem spurred the Tokyo- This approach provides users with a more efficient way of
based pharmaceutical giant to begin a journey toward a zero navigating Takeda’s complex technology environment—a
trust-based security architecture. mix of cloud-based applications and services and legacy
systems residing on internal servers and in data centers—
“We realized that the demarcation between internal and and eliminates the difficulty of accessing systems via multiple
external was no longer relevant or scalable,” says chief firewalls and VPNs. The transition to zero trust, well underway
information security officer (CISO) Mike Towers. “The zero before COVID-19 struck, helped the company securely
trust mindset—assuming that every request to connect is manage the sudden shift of its global workforce to a work-
coming from an unknown access device on the internet that from-home model. “Our China workforce, the first affected by
can’t be predicted or controlled—is a much better way to the pandemic, had less experience and comfort with work-
move forward.”5 from-home, so it was really important for us to get it right,”
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 108
Towers says. “By having shifted to zero trust- the organization gets a solid understanding of suppliers began pushing their products and
based access, we were able to aggressively their application landscape.” services to the cloud. “With the dispersion of
and quickly move China to the work-from- our computing resources from the data center
home model.” Towers believes that once leaders understand to the internet, we realized that our traditional
the numerous benefits, most companies network perimeters were dissolving,” says
Setting clear expectations with business will inevitably adopt the zero trust mindset. Mary Rose Martinez, CISO and senior director
partners is critical during the transition, “Frankly,” he says, “I don’t think that businesses for IT architecture.6 “This impelled us to
says Thomas Likas, global head of security can digitally or technologically scale in any develop a zero trust strategy.”
architecture and engineering. He recommends other way.”
that security and IT organizations planning Halliburton’s zero trust approach revolves
a zero trust migration engage with business
partners from the beginning of the journey.
Zero trust secures the around securing people, network connections,
and data. “We are moving toward a reality
“The business—not IT—has the best “new perimeter” where it doesn’t really matter if employees are
understanding of how people access and use on the network or not,” Martinez says. “The
their applications,” he says. “In the zero trust A zero trust approach is helping Halliburton, a new perimeter is identity, whether user identity,
world, the business will need to determine global provider of products and services to the endpoint device identity, or service identity.”
who should have access to their systems energy industry, meet its strategic business
and data.” goals and objectives. Several years ago, as part When Halliburton began its zero trust journey
of a drive to be more operationally efficient, about two years ago, it focused first on
Indeed, Likas continues, “this knowledge the company began adopting cloud, mobile, securing mobile devices through multifactor
needs to be baked into the access model from and Industrial Internet of Things platforms authentication—using identity credentials,
the very beginning. To business partners, this to reduce costs and improve productivity. an authenticator, and registered devices.
might seem like a lot of work, but as a bonus, At the same time, Halliburton’s vendors and Soon after, the company migrated to cloud-
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 109
based identity providers to further secure its network or the internet, in the office or at
people. Over time, the number of applications home, they receive a verification prompt before
accessible without using a VPN continues to accessing applications protected by multifactor
grow. A longtime adopter of the principle of authentication. This workflow change required
least privilege, data encryption, and other education and awareness. And it is incumbent
data controls, the company is also working to on users to guard their credentials and devices.
enhance the classification and protection of “Raising security awareness has to be part and
unstructured data. parcel of the zero trust approach,” Martinez says.
“An ongoing education program that includes
The more granular security controls that are increasingly sophisticated phishing simulations
part of Halliburton’s zero trust approach have can help people become more aware.”
created a more disciplined security posture.
Because it controls user devices and endpoints, Halliburton’s adoption of zero trust is an
the company can push policies to any device ongoing journey, with many components that
via the internet. And because VPN access isn’t are constantly moving and changing shape.
required for the zero trust-enabled applications, “Because of the fluid nature of technological
employees have a considerably improved advancements, the end state will probably
user experience. always be a moving target,” Martinez says. “But
we’ve laid a foundation that’s both solid and
Martinez is quick to emphasize that zero adaptable, and upon which we can continue to
trust is not only a technology initiative—it is build over time.”
also a people initiative. For example, whether
Halliburton employees are on the company
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 110
bad intent are often in positions of trust. This a product, although zero trust-based security Zero trust is a journey best taken one step
means it’s impossible to know with certainty infrastructures can be implemented by using at a time. I recommend that organizations
that the originator of network traffic can truly many different products. Nor does zero trust begin by prioritizing the smallest possible
be trusted: An asserted identity is only an require organizations to rip and replace existing protect surfaces—a single data set, asset,
assertion, not an actual person. security infrastructure—rather, it leverages application, or service—depending on the
existing technology to support the zero trust level of sensitivity or business criticality. Then,
In response to what CISOs and CIOs told me mindset, with new tools added as needed. they can create a microperimeter around each
about their cybersecurity strategies, I created protect surface and granularly control the
the concept of zero trust, which is framed traffic allowed into the perimeter.
around the principle that no network user, The hallmark of zero trust
packet, interface, or device—whether internal
or external to the network—should be trusted. is simplicity. I encourage security teams to learn and
practice on less sensitive protect surfaces,
Some people mistakenly think zero trust is moving to protect increasingly more sensitive
about making a system trusted, but it really The hallmark of zero trust is simplicity. or valuable protect surfaces as they fine-
involves eliminating the concept of trust from When every user, packet, network interface, tune their approaches and their confidence
cybersecurity strategy. By doing this, every and device is untrusted, protecting assets increases. Over time and with lots of practice,
user, packet, network interface, and device is becomes simple. To reduce the complexity of they’ll be ready to migrate their most critical
granted the same default trust level: zero. cybersecurity environments, organizations can assets to the zero trust environment. Finally,
prioritize security technologies and tools that when high-value assets are protected, teams
Zero trust should be thought of as a strategy support simplicity by automating repetitive can focus on less important assets. And by
or framework. It requires companies to rethink and manual tasks, integrating and managing continuing to maintain a zero trust mindset,
their philosophy and approach to trusted multiple security tools and systems, and organizations can protect themselves even as
network users and devices. Zero trust is not autoremediating known vulnerabilities. security technologies and tools evolve.
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 112
EXECUTIVE PERSPECTIVES
STRATEGY // The stakes are high for FINANCE // Reporting on cybersecurity RISK // Zero trust is fast becoming
CEOs when it comes to cyber risk. Beyond breaches is among the CFO’s more the modern standard for managing
the damage that security breaches can unpleasant responsibilities. When these infrastructure, network, and data in a
have on companies, shareholders, and customers, events happen, CFOs are often on point, reporting more secure manner. Despite the concept’s broad
they can end careers. CEOs are often answerable to to auditors and answering to the board, regulators, benefits, many see it as solely a technology issue. To
the public for their organization’s security posture, and the public. In a time when cyber risk is increasing change that, over the next 18–24 months, CROs should
especially as it relates to consumer data, and they and bad actors regularly test organizational defenses, consider placing zero trust adoption at the top of their
should thus approach this topic as the stewards CFOs should develop and maintain technology agendas. CROs can first clarify the security benefits
of the organization’s brand reputation and trust. fluency and the awareness they will need to to the organization and then work with the CIO, CISO,
By appointing a CRO, CISO, or other appropriate mitigate cyber events. Moreover, they should clearly and other leaders to enforce the new approach.
leader, they can look to new security postures such understand the risks and rewards of their company’s Thorough adoption can eventually help risk posture
as zero trust that simplify protection of data, people, security posture—particularly as it applies to key and processes evolve in lockstep with innovation while
and networks without sacrificing user experience. strategic, physical, and financial assets—and then reducing the frequency of cyber incidents.
Setting security priorities from the top can help the improve security by enabling zero trust adoption.
organization align on the importance of strengthening Ultimately, the CFO—working in tandem with other
cyber defenses. risk and security leaders—can become a de facto
crisis manager, working to predict and prevent threats
to brand reputation, shareholder trust, and more.
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 113
1
How far are you on your journey Explore the latest insights on managing
moving away from network and cyber risk and bolstering security
server “zones of trust”? What is against attacks.
your next step?
2
skills and capacity of your imperative in uncertain times
cybersecurity organization relative Read a joint cybersecurity report
to today’s challenges? What about with the National Association of
tomorrow’s? State Chief Information Officers.
3
business-function system owners Tune into an award-winning podcast
in security planning? Would their series that features leaders tackling
help in identifying areas requiring risk, crisis, and disruption.
more restricted access improve
the overall security posture?
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 114
AUTHORS
Our insights can help you take advantage of emerging trends. If you’re looking for SENIOR CON T R IBU TOR
fresh ideas to address your challenges, let’s talk.
Wil Rockall
Deborah Golden Arun Perinkolam Partner, Deloitte MCS Limited
US Cyber & Strategic Risk leader Cyber & Strategic Risk core
Deloitte & Touche LLP infrastructure and network
offering leader
debgolden@deloitte.com
Deloitte & Touche LLP
aperinkolam@deloitte.com
Mark Nicholson
Cyber & Strategic Risk marketplace
development leader
Andrew Rafla
Deloitte & Touche LLP Cyber & Strategic Risk zero trust
offering leader
manicholson@deloitte.com
Deloitte & Touche LLP
arafla@deloitte.com
Kieran Norton
Cyber & Strategic Risk infrastructure
security solution leader
Deloitte & Touche LLP
kinorton@deloitte.com
1 2 3 4 5 6 7 8 9 Zero trust: Never trust, always verify 115
ENDNOTES
1. Business Wire, “Opus & Ponemon Institute 6. Mary Rose Martinez (CISO and senior director for
announce results of 2018 third-party data IT architecture, Halliburton), phone interview with
risk study: 59% of companies experienced a the authors, October 2, 2020.
third-party data breach, yet only 16% say they
effectively mitigate third-party risks,” November
15, 2018.
The digital workplace’s technologies and tools Data and insights from the digital workplace can
can help employers gain insights that enable help organizations improve work processes and
them to identify patterns, make predictions, employee experience to create better outcomes.
nudge behaviors, and fine-tune performance.
TREND 7
D
riven by COVID-19, the abrupt shift to telecommute more after pandemic restrictions • Relationship building and onboarding.
remote work as the default for much are eased.2 Strong existing relationships typically
of the labor force has altered work as remain so after a remote work transition,
we know it. One study found that only 15% of As organizations manage a growing number but weak ones often erode. And it can
those employed pre–COVID-19 worked from of offsite employees—working from home or be challenging to virtually cultivate new
home; these workers were joined after the other nontraditional workspaces—many are relationships and onboard new employees.
pandemic by an additional 35%, suggesting beginning to accept the inevitability of a digital
that fully half of the employed labor force now workplace where work is completed by a mix of • Development and learning. It could be
works from home.1 onsite and remote workers that must operate difficult for workers to virtually develop and
in synchrony to meet business objectives. maintain skills and professional networks
With the work-from-home cat out of the Some leaders approach the prospect of the they need to advance.
bag, it could prove challenging to coax these digital workplace with a number of concerns:
professionals back to prepandemic levels • Impact on innovation. The loss of
of in-office work. Staffing firm Robert Half • Productivity. Distractions and poor serendipity—the unexpected and
found that 60% of professionals who had supervision could impede collaboration opportune employee interactions that fuel
transitioned to a remote setup said they and derail productivity, and the lack of exploration and invention—could have a
have better work/life balance, and that 74% social support might lead to high employee long-term negative impact on innovation.
of respondents would like to continue to dissatisfaction and turnover.
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 118
Companies may be able to overcome the tools can help employers gain insights ranging environments and services that support secure
digital workplace’s deficits and ambiguities from individual employee performance to remote access and scale with flexing workloads.
by more intentionally embracing its positive team-level productivity to companywide Of Deloitte’s own experience in rapidly
aspects, including the data generated by morale, enabling them to identify patterns and enabling its workforce, unified communications
workers’ tools and platforms. This can make predictions, nudge positive employee infrastructure lead Aaron Roe notes that
help organizations optimize individual and behaviors, and fine-tune individual, team, and previous infrastructure upgrades proved their
team performance and customize the organizational performance. value. “Because we had already moved to
employee experience through personalized cloud-based unified communications, we were
recommendations, enabling remote work to Enabling technologies include people analytics able to migrate the entire firm to remote work
be far more than a diminished proxy for the solutions that help deliver actionable insights on in less than four days without overloading our
traditional office. And as onsite workspaces employee and team behavior and productivity, data centers,” he says. “And our move to a
and headquarters evolve, organizations can AI-driven personal productivity assistants and zero trust security model allowed employees
use this data to create thriving, productive, other artificial intelligence (AI) technologies to securely access needed resources without
and cost-effective offices that are seamlessly that use workforce data to make personalized using a VPN.”3
interwoven with the remote experience. recommendations to optimize performance, and
workflow management solutions that streamline On top of these foundational technologies,
The most agile companies will investigate workplace, managers can identify employees could offer employees assignments predicted
the patterns remote workers are encoding in suffering from emotional stress and burnout to be both interesting and skills-aligned. The
data and use them to develop new ways of and proactively intervene to address such coach “knows” the skills and experiences
working. As ways of working and collaborating challenges as low engagement and lack that each employee needs to be optimally
are standardized, the knowledge that can be of inclusivity.6 Psychographic data such as challenged and makes suggestions to improve
extracted from emerging work patterns will boredom, stress, and fatigue levels can help behavior, collaboration, and specialized skills in
become more precise and more valuable. leaders better allocate tasks and ensure that real time.
employees are appropriately challenged and
and peers. Gig platforms, or internal talent For example, even though employees desire could find that creating exciting environments
marketplaces, can help connect employees the flexibility to work remotely, most aren’t that employees are eager to experience
with internal and external opportunities, ready for an online HQ. In one survey, nearly is the best way to entice them back to the
mentorships, stretch assignments, and three-quarters of workers said they prefer a office. Perhaps, as architecture and design
rotation programs and enable managers to balance of working at home and in the office.7 firm Gensler predicts, the post–COVID-19
find needed skills. The primary draw of the office, according workplace will shift from a place where people
to another study, is face-to-face human work to a place where teams meet, socialize,
connection,8 particularly for millennials and and connect.10
The office is dead, long live Gen Z workers.9
The smart money is on flexible and
the office!
While the office
configurable work environments, technology-
driven workplace services, and new ways
While the office may be undergoing a
pandemic-driven existential crisis, it’s not
may be undergoing of managing fewer people and less space
without sacrificing collaboration and
down for the count. (The everyday commute,
on the other hand, appears to be on life
a pandemic-driven innovation. As the office evolves to become
both a collaboration hub for project teams
support.) However, just as e-commerce existential crisis, it’s and a creative center for client meetings,
changed the way consumers and retailers employees can expect a hybrid work style
think about brick-and-mortar storefronts, so not down for the count. that supplements working from home with
is the forced mass adoption of remote work targeted in-office experiences, especially
changing the way employers and employees How do we reconcile these competing needs? for critical events such as onboarding. And
think about the physical workplace. The office may not be dead, but it’s unlikely to while coworking spaces took an immediate
return in its previous incarnation. Employers occupancy hit when COVID-19 first struck, in
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 122
the long term, they could be a more flexible, optics and noise-canceling microphones at considering how to configure and manage a
less expensive office alternative for businesses each seat to provide remote attendees with a digital workplace in which work is completed
that are canceling or renegotiating their leases. more realistic experience, serves as a model by both remote and onsite workers. The data
for how such parity might be achieved. generated by workers’ tools and platforms
Offices of the future will likely be infused with can help organizations fine-tune individual
the same digital technologies and tools used in Organizational AI can help teams organize productivity and team performance, deliver
the remote workspace. The same capabilities the dates, times, and locations of in-office customized employee experiences, and
that enable a personalized employee meetings to maximize the value of space optimize the use of office space. As employees
experience in the digital workplace can similarly and promote team interactions. And with return to the office, this data can also help
augment the physical workplace experience. good office design, companies can measure remote and onsite teams work in concert
the impact of traditional office orthodoxies and ensure the parity of remote and in-office
The office will likely be retooled to more easily such as in-person conversations, water- employee experiences.
permit virtual, multidirectional collaboration cooler discussions, and spontaneous hallway
with remote workers. For example, in Deloitte encounters that generate real business value.
offices, all conference rooms, phone rooms, Yet the return on investment for the office
and other closed-door meeting spaces will of the future will likely be more carefully
be enabled for video with platform-agnostic examined than it has in the past.
videoconferencing solutions that integrate all
leading applications. Teams are exploring how
technologies can create parity of experience The way forward
between people meeting in the office and
remote participants. Deloitte’s boardroom, In the wake of the mass forced adoption
featuring individual cameras with advanced of remote work, many organizations are
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 123
LESSONS
to the home workplace. The use of videoconferencing tools
New tools, good timing support became the norm for daily meetings while the use of desk
FROM THE remote work rollout phones faded. In many cultures, a deep-rooted sense of
formality toward behavior in the work environment made
FRONT LINES New tools and good timing helped Thomson Reuters navigate
the pandemic-driven shift to remote work, says Rob Newnes-
some employees feel awkward about the casualness of
working from home. Team leads cultivated authenticity and
Smith, CTO of digital, employees, and partners for the compassion by modeling more relaxed attire and the use of
business information services and global media company.11 home areas not typically seen as office space, with perceived
When the majority of its 25,000 employees began working erosion of boundaries such as videoconference interruptions
from their homes in March 2020, Thomson Reuters was by children or pets.
putting the finishing touches on a major global rollout of
cloud-based enterprise systems and platforms, including Having successfully enabled its employees to be productive
collaboration and productivity tools that made it easier for in the remote work environment, Thomson Reuters is
employees to quickly adapt to remote workflows and working to define the postpandemic workplace experience.
maintain productivity. Its well-timed modernization of enterprise systems puts the
company in a good position to adapt to “the next normal”
A significant portion of Thomson Reuters employees were with access to personalized, consumable information,
accustomed to periodic virtual work, but these previous improved use of automation and AI to complement self-
experiences provided only partial insights into the impact service functionality, and richer levels of insights available
of long-term remote work and lockdown isolation on well- for the executive leadership team. Newnes-Smith envisions
being, especially when other factors such as home-schooling employees routinely dividing their time between regional hubs
and care for elderly relatives were in play. Company leaders and their home offices so that they can enjoy the best of both
recognized the importance of extending the corporate culture worlds: face-to-face collaboration, socializing, and unplanned
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 124
interactions along with the flexibility of Reuters continues to hone its use of various It was a dramatic change for participants in
telecommuting to create an improved work/ collaboration tools to ensure employees’ the insurance market, which had never in its
life balance. voices are heard and faces are seen. “When 330-year history halted in-person trading.12
everyone is an equal square on a screen,” Traditionally, the four-story underwriting
Moving forward, the Thomson Reuters HR, real Newnes-Smith says, “it has a way of leveling room is a bustling trading hub packed with
estate, and technology teams are deliberately the interaction, whether you’re located at insurance brokers and underwriters who
reenvisioning in-office technologies, including home, in an airport, or in the office.” conduct face-to-face meetings to negotiate
integrated communication and collaboration deals between policy buyers and sellers. On a
Lloyd’s of London
platforms, new tools such as smart usual day, the underwriting room hosts about
whiteboard technology, idea management 5,000 market participants; around 45,000
software to crowdsource innovative ideas accelerates innovation with people have badges to enter. “Historically, the
from employees, predictive talent insights, market floor is the heart of Lloyd’s. Business is
automation of key talent acquisition base virtual underwriting room grounded in the physical space, where brokers
processes, and AI-based tooling as part of walk the market floor and wait in queues to
a new learning system platform. “We plan When the COVID-19 pandemic hit the United see which underwriters are at their booths
to leverage our collaboration platforms to Kingdom, Lloyd’s of London—the renowned and whether they might be interested in
create a frictionless, seamless experience as commercial insurance marketplace that insuring a particular risk,” says product owner
employees move between their chosen places intermediates between clients, underwriters, Sam Irving. “COVID-19 was a monumental
of work,” he says. brokers, and insurance companies—closed its event because it completely disrupted these
market floor, known as the underwriting room. well-established workflows.”13
Ensuring equity between in-office and at- From their hastily set up home offices, brokers
home work experiences is critical. As we move and underwriters transitioned from in-person As Lloyd’s worked toward a phased September
into the ninth month of COVID-19, Thomson deal-making to electronic trading. reopening, leaders looked to leverage digital
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 125
technologies to make connecting more that centered the needs of brokers and and supportive experiences whether the user
operationally resilient, instead of simply underwriters—Lloyd’s established a virtual is trading remotely or on the market floor.
returning to business-as-usual in the physical room that connects brokers and underwriters For example, the market directory improves
underwriting room. “This wasn’t a completely via digital collaboration platforms, enabling the efficiency of brokers in the physical
new idea for us, but the pandemic’s challenge them to schedule trading conversations with underwriting room by enabling them to check
to the old ways of working really accelerated colleagues in different locations around the underwriters’ availability without walking
our thinking,” Irving says. world. The virtual room includes searchable around the market floor, and as market
broker and underwriter profiles and a market profiles develop, this could help brokers
Leaders decided to develop a virtual directory, availability settings, and flexible chat confirm underwriter interest in potential deals
underwriting room that would complement functionality. This allows brokers to search for without standing in queue.
Lloyd’s’ in-person counterpart. This meant underwriters by specialty, view underwriter
identifying gaps in the existing electronic profiles, see who’s free for an immediate chat, An unintended benefit of the virtual room
trading experience, such as the ability to and schedule meetings or videoconferences. is that publishing users’ availability can aid
schedule on-the-fly meetings, ask and answer The market directory and profiles help their work/life balance. “Working from home
quick questions, and meet spontaneously— underwriters promote themselves by removed the traditional end-of-the-day
the hallmarks of the market floor. The goal providing more visibility into the types of boundary of physically leaving the market
was not only to improve the current work- opportunities they’re seeking. floor and going home, and people routinely
from-home experience but to leverage those contacted each other on their mobile phones
learnings to improve the traditional in-person Lloyd’s unveiled its virtual underwriting during nonwork hours,” Irving explains.
trading process. room to selected users in tandem with the “Setting their availability gives remote users
September reopening of the market floor. The more control over developing and maintaining
On an accelerated timeline—and backed by goal is for the virtual and physical rooms to work/life boundaries.”
an extensive research and design process work in synergy and provide equally efficient
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 126
Lloyd’s continues to build on the initial technologies, expects the road to the “next feel supported waned during the pandemic,”
launch of the virtual underwriting room, normal” to be fluid and complex. Puybaraud says. “Therefore, we don’t expect
adding and improving features and releasing office environments to retain traditional
improvements weekly. Top priorities for the COVID-19 has accelerated the transformation designs. We have a flexible workforce, and we
future include virtual queuing, integrated of the workplace by five to 10 years, says will need workplaces to be equally adaptable.”
calendars, and mobile applications. Marie Puybaraud, JLL’s global head of
corporate research, who specializes in For example, instead of a single corporate HQ,
a third set aside for collaborative purposes. To The success of the mass work-from-home
meet the needs of the future workforce, these experiment conclusively proved the value
proportions likely will need to flip, with roughly of technology, Puybaraud says. “Technology
three-quarters of the workspace for collective established itself as a successful enabler of
use and the rest set aside for individual work. remote work. By and large, businesses were
able to continue to operate efficiently because
This likely will require businesses to of technology platforms. However, with
significantly retool and redesign their office employees dispersed among multiple office
environments to meet workforce preferences locations and home offices, companies will
and reenvision the workplace as a social hub probably need to expand those platforms as
to boost human experience and performance. we move forward. There is a lot of opportunity
JLL is exploring how artificial intelligence, for new solutions that deliver more realistic,
virtual and artificial reality, 3D modeling, engaging, human-centric experiences. While
and other technologies can help push the nothing can replace the real-world interactions
built environment beyond its current limits. that most people crave, technology can make
Besides transforming the physical workspace, interactions between on-site and remote
technology can enable the dispersed colleagues even more efficient.”
workforce to collaborate and innovate
effectively. For example, future technology
platforms need to enable those who aren’t
physically present in future collaboration
spaces to interact naturally and productively
with those who are.
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 128
and data centers, PayPal But as it became clear that the workforce would remain offsite for
the foreseeable future, we shifted our focus, using the rapid move
to remote work as an opportunity to reexamine the ways we work
together in teams and as a company.
challenges of working away from the inherently human challenges as sustaining With most or all employees operating from
corporate office. social connections with colleagues, managing their homes, it’s incumbent upon employers
family members during work hours, and to optimize remote worker experiences,
Before the pandemic, work processes, maintaining work/life balance. The digital intentionally evolving them to be as efficient
collaboration tools, communication channels, workplace requires employers to reengineer and delightful as onsite experiences are.
learning opportunities, and culture—even the way their people connect, communicate, Moving forward, for example, we’re looking for
the way we socialized with colleagues—were collaborate, and support each other within ways to tailor the digital collaboration process:
necessarily optimized for the traditional and across various functional areas. We’re exploring how to deliver the same
workplace. For example, about half of our capabilities and experiences across multiple
employees work in operational roles in
customer service, collection, and fraud.
With most or all employees collaboration platforms so that each
employee can work with others using their
Before the pandemic, they worked in highly
controlled, metrics-oriented call centers;
operating from their preferred tools.
their colleagues were all within arm’s reach. homes, it’s incumbent Employers can also identify and nurture local
COVID-19 required us to evaluate the and regional working norms, differences in
sustainability of such traditionally designed upon employers to optimize IT infrastructure, and technology preferences
teams. Now, these same employees are
working solo—or with roommates, spouses,
remote worker experiences, that could affect employee work styles and
processes. For instance, bandwidth
and children—in unstructured home-office
environments. They’re still meeting their
intentionally evolving them to capacity, internet support, and network
capabilities, along with cultural differences
productivity goals, but with new challenges be as efficient and delightful in attitudes toward working from home, vary
related to communicating, collaborating, among regions. Understanding and
skill-building, and networking as well as such as onsite experiences are. adapting to such factors can help companies
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 130
EXECUTIVE PERSPECTIVES
STRATEGY // For many CEOs, this year’s FINANCE // The challenges of the rapid- RISK // This year’s jarring shift from
comprehensive shift to remote work onset digital workplace continue to be top onsite to remote work offers CROs an
remains top of mind. Leaders who are of mind throughout the C-suite and broader opportunity to redefine security standards
eager to improve the work experience can look to organization. CFOs should consider the potential for remote collaboration. For example, home
new digital tools that monitor online interactions to financial impacts of these challenges, including the networks now process the same level of workloads as
explore short-term improvements such as better downsizing of large, urban offices into dynamic, smaller corporate networks, and the increase in VPN usage
collaboration as well as longer-term innovation. The workspaces. But they should also anticipate potential means increased commingling of company data
virtual office also offers an opportunity for leaders costs of diminished productivity and creativity, as and personal devices. As digital collaboration and
to practice and promote radical transparency employees no longer congregate physically. In virtual file-sharing become routine, risk leaders should try
in communication. Still, CEOs should be highly settings, we often lose the ability to read social and facial to respond constructively to the increasing pace of
purposeful in architecting their goals for remote cues, or to brainstorm during informal, water-cooler change without hampering the organization’s ability
work, as the choices they make can affect company conversations. Finance leaders can lead differently in to keep pace with the market. They can work closely
culture well beyond the pandemic. They can either remote work by setting a strategy to address some of with the CIO and CTO to assess upcoming changes
help employees feel more connected than ever to these gaps in the new workplace—for example, they to collaboration tools and decide on a risk posture
each other and to the company’s vision or leave them can implement AI tools to analyze meetings in real time that works for the near term. Risk leaders can also
feeling disconnected and alone in the new normal. and nudge decision-makers to solicit opinions from anticipate further challenges and changes as a broad
quieter participants.15 Whether or not companies adopt return to corporate offices takes place.
such technology measures, people will look to CFOs to
find ways to support employees’ well-being and their
ability to deliver outcomes that sustain the company.
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 132
1
How do you assess employee work infrastructure
and team performance of remote Study the cloud solutions, security, and
workers? What steps are you DevOps methodology that can make
taking to improve performance? virtual business infrastructure work.
2
What steps have you taken to Survey the latest changes in store for
virtually nurture the spontaneous the workplace, workforce, and the
employee connections and nature of work itself.
dialogue that drive innovation?
3
the long-term role of physical Navigate the future of human-
workspace? How will office design machine collaboration in the
and technology tools support workforce as AI adoption grows.
optimal collaboration and
productivity?
1 2 3 4 5 6 7 8 9 Rebooting the digital workplace 133
AUTHORS
Our insights can help you take advantage of emerging trends. SENIOR CON T R IBU TOR S
If you’re looking for fresh ideas to address your challenges,
let’s talk. Jeff Schwartz
Principal,
Steven Hatfield Deloitte Consulting LLP
Global Future of Work leader
Yasar Butt
Deloitte Consulting LLP
Director,
sthatfield@deloitte.com
Deloitte MCS Limited
ENDNOTES
1. Erik Brynjolfsson et al., “COVID-19 and remote create a better way to work,” Slack, September 1, 15. Sandy Pentland, “Context is key when leading
work: An early look at U.S. data,” NBER working 2020. virtually,” Deloitte CFO Journal, April 20, 2020.
paper 27344, June 2020. 8. Andy Cohen and Diane Hoskins, “Insights from
2. Robert Half, “Survey: Employees share views on Gensler’s U.S. Work from Home Survey 2020,”
current and post-pandemic workplace,” news Gensler, May 26, 2020.
release, May 1, 2020. 9. Gensler, “Younger generations are less productive
3. Aaron Roe (unified communications infrastructure at home—and overall, less satisfied with the work
lead, Deloitte), phone interview with authors, from home experience,” 2020.
November 4, 2020.
10. Janet Pogue McLaurin, “Most people want to
4. Susanna Ray, “Video fatigue and a late-night return to the office—but they expect changes,”
host with no audience inspire a new way to Gensler, May 22, 2020.
help people feel together, remotely,” Microsoft 11. Rob A. Newnes-Smith (CTO of digital, employees,
Innovation Stories, July 8, 2020. and partners, Thomson Reuters), phone interview
5. Allan V. Cook and Kaitlyn Kuczer, A brave new world with authors, October 2, 2020.
with virtual worlds: How virtual worlds are the next 12. Reuters, “Lloyd’s of London considers part-virtual
evolution in communication and collaboration, underwriting room,” May 29, 2020.
Deloitte Insights, October 2, 2020.
13. Sam Irving (product owner, Lloyd’s of London),
6. Erica Volini et al., Designing work for well-being: phone interview with authors, September 30, 2020.
Living and performing at your best, Deloitte
14. Marie Puybaraud, PhD (global head of corporate
Insights, May 15, 2020.
research, JLL), phone interview with authors,
7. Brian Elliott, “Slack launches the Future Forum to November 2, 2020.
135
TREND 8
D
uring the COVID-19 pandemic, a tailor In what we recognize as an emerging trend, transactions, or something else. These digital
on London’s Savile Row—unable to consumers are becoming less satisfied alternatives to in-person activities initially
travel to customers—found himself with distinct physical or digital experiences. seemed adequate, but the law of diminishing
measuring clients two continents away in They want the best of both: personalized returns quickly asserted itself.3 As it turns
Seoul for bespoke suits ... with the assistance interactions combined with the convenience out, many of the digital experiences that have
of a robot.1 By necessity or choice, people of digital. During the next 18 to 24 months, become mission-critical leave customers
have increasingly embraced digital interactions we expect to see leading companies embrace wanting more, which is hardly a recipe for
in all aspects of their everyday lives, whether the bespoke for billions trend by exploring ways long-term success.
in working remotely, online schooling, or to use human-centered design and digital
What does a bespoke digital experience look Savile Row’s challenge and opportunity is no longer be needed as an interface: The
like for your customers? They’re looking for to retain the cachet of customization while consumer will get the experience delivered
an answer. widening availability, with the aid of various when, where, and how they want it without
technologies. Leading companies and brands even requesting it. As an early example, for
Physical + digital is the are looking to arrive at the same place— years ride-share companies have used ML
from the opposite direction. Many are in the and data-driven algorithms to predict where
new bespoke early stages of experimenting with existing and when consumers might need rides.4 As a
and emerging technologies, combining result, the ride-share car is often right around
Humans have long valued experiences and them in innovative ways to design and scale the corner when you exit the building, before
products that are customized for them. more personalized human experiences and you even open the app.
Since the early 1800s, London tailors have outcomes. As opposed to mass customization,
created bespoke experiences working in which consumers can choose from a limited Large gaps remain in many physical-plus-
intimately with clients who select fabrics, number of options, a bespoke experience is digital consumer experiences. Let’s say
buttons, and patterns to compose a physical tailored specifically to an individual’s behaviors, a couple downloads an online retailer’s
expression—a flattering one, naturally— preferences, values, and beliefs. Eventually, we augmented reality (AR) app that allows them
of their taste, personality, and ambitions. expect that technological advances will allow to envision a sofa in various styles, fabrics,
Traditionally, a tailor might create custom companies and brands to create bespoke and colors in their living room. While digital
clothes for only dozens of regular customers experiences for billions of people—not just experience can suggest whether and how the
during a career. In a field long based on the privileged. size and design will complement their living
hands-on personal attention rather than space, the experience is incomplete, since
technology and automation, comparatively Right now, there is typically a smartphone the customers are unable to experience the
few have experienced creating and wearing or tablet between the consumer and the sofa’s physical comfort. Indeed, because many
bespoke clothing. experience. But over time, the device will consumers are still reluctant to purchase
1 2 3 4 5 6 7 8 9 Bespoke for billions: Digital meets physical 138
without physically experiencing items— with curated items based on that individual’s Over time, we expect, digital interactions
especially expensive ones—some online online behaviors and preferences.6 will enhance a growing number of physical
retailers are opening physical stores to round experiences, and vice versa. The
out the shopping experience. For example, Imagine that consumers could opt in to convenience and simplicity of digital
Casper, an online mattress company, has continue that journey in the physical world interactions will eventually be seamlessly
opened 20 stores, with plans for more, to with a personal experience tailored to their infused with the warmth and assurance of
allow customers to try out their mattresses; individual tastes and preferences. One meaningful human exchanges.
they can even book a nap.5 consumer may like a dress she sees online but
hates dealing with returns; she wants to try
Without a digital infrastructure in place, sectors the dress on before deciding to buy. Perhaps Design for an audience of one
traditionally grounded in the physical world this customer requests a reminder when
will likely be slower to evolve. To overcome she finds herself in the vicinity of a branch In a January 2020 survey, 3,000 C-suite
this challenge, some brick-and-mortar of the department store. If she prefers not executives said that they believe elevating
companies are partnering with digital natives to interact with a salesperson, her mobile the human experience should be a top
to blend their physical human experiences device—equipped to access the spatial web— organizational priority, yet 96% of them have
with digital elements. For example, one social could help her locate the dress in the store. struggled to design and launch anything
media platform offers its retail advertisers a Another individual may prefer that the store’s resembling human-centered experiences.7
customized online storefront that delivers a virtual stylist arrange to have the dress, along
seamless digital experience to attract new with coordinating accessories in her preferred Pioneering companies and brands are
customers from the platform’s vast social colors and sizes, waiting for her in a private exploring ways to fill this need with human
network. Targeted social ads entice a consumer dressing room. And yet another individual may experience designers who create value-
to click on desirable items to learn more, which simply want a personalized selection of clothing enriched human interactions, both physical
takes the consumer to a mini website stocked delivered to her home. and digital. These designers possess the
1 2 3 4 5 6 7 8 9 Bespoke for billions: Digital meets physical 139
sensibilities, empathy, and imagination could become obsolete. There’s a new breed to tweak the experience to suit that person,
needed to create satisfying, trust-enhancing of human experience creative: designers who essentially creating bespoke experiences
interactions for personas, which are groups use code as their medium, eliminating the gap based on understanding who the individual is,
of consumers who share similar needs, between design and execution. By combining where they are, and which offers they are likely
behaviors, and goals. They create highly creative sensibilities with technical know-how, to want.
nuanced elements of the customer journey these designers/coders can create increasingly
and specify the various ways those elements
can be strung together to reach an outcome
seamless experiences based on individuals’
behaviors and preferences, defined by real-
Technologies scale trust
that’s satisfying for both the customers and time data that’s relevant in the moment. for billions
the brand. Experiences designed for groups
one experiences to fit billions? Accomplishing behave in trustworthy and ethical ways, as The figure “Technology tools of the bespoke
this will require thoughtful human design we described in Ethical technology and trust experiences trade” (page 141) outlines the
amplified with trust-enabling technologies. in Tech Trends 2020. Moreover, meeting the foundational technologies supporting the
Two main technologies that can enhance— local, regional, national, and transnational bespoke for billions trend; many of these have
or destroy—trust are AI and the use of patchwork of data and data-usage regulations been around for years, becoming more
personal data. will require careful planning and engineering. sophisticated with each evolution. What’s new
are the possibilities that are created at their
Design, Gensler
they already have a robust digital infrastructure in place.
physical buildings and interiors to creating revolutionary to a digital native, but it is for an that built environment is part of a greater
holistic experiences. We are merging physical analog industry like real estate. In a way, journey. So, as the team owner, you have a
spaces and digital interactions to create we’ve moved ahead seven to 10 years in just vested interest in not just enriching a fan’s
personalized experiences for the end users, six months. interactions in the stadium—you also want to
whether they are employees, sports fans, understand how to connect and engage
students, passengers, or shoppers. with your fan before, during, and after the
We are helping Gensler game. We’re considering the full 360-degree
the way to the exit or the elevator. In the One big challenge that’s holding us back from
new world, when a remote worker visits the Gensler has also broadened our design reach offering more personalized experiences is the
corporate office, the meeting invitation on to include the user’s experience before, lack of an “urban platform,” a common chassis
his computer tablet is the clearing pass for during, and after visiting a building. For for building customized digital elements for
security, the elevator key, and the map to example, a soccer fan doesn’t engage with the physical world of architecture. We’re
the conference room. That may not sound the team only when they’re in the stadium— actively working on that but until that’s fully
1 2 3 4 5 6 7 8 9 Bespoke for billions: Digital meets physical 145
evolved, designing and building 360-degree and we keep investing in the next generation.
user experiences will continue to be slower We’re very much driving our own future.
and more expensive than we and some clients
would like.
EXECUTIVE PERSPECTIVES
STRATEGY // The pandemic has given FINANCE // As human-centered design RISK // Physical and digital are blending
fresh importance to the CEO’s role as an improves and IT brings more use cases to create hybrid user experiences. In this
end-user ethnographer.12 Organizations to the finance team, CFOs should be new environment, privacy-related issues
are scrambling to respond to customers’ values, mindful of the business case: Which metrics (ROI, may be amplified relative to purely physical or digital
risk appetites, and preferences as they design safe cost per impression, etc.) will justify investments, and customer interactions. This shift can lead to CROs
experiences that blend the physical and digital. CEOs which products or services are ripe for disruption? rethinking traditional forms of security such as the
are on point for directing this human experience Moreover, in a primarily digital economy, CFOs should password and opting for newer methods such as
strategy. They should consider shifting trends on consider how human connections drive accounting mobile biometrics. They should pay close attention
what technology is enabling (for example, increases in and valuation. They may need to revisit concepts to privacy as technology enables organizations to
virtual shopping and service) as well as the differences of impairment or sales tax implications to assess increasingly identify unique customers and their
in what people are expecting from brands. Ultimately, whether such standards preclude them from showing data across channels. Risk leaders may also be
a consistent, humane experience is the goal, and strong financial statements. Likewise, the value of dealing with a generational divide on human
setting direction from the top down can help CEOs high-quality human experience platforms may be experience, as expectations vary widely between
lead their organizations in line with their values—and difficult to quantify, but it can also become a leading digital natives and older generations.13 Leaps in
their vision for responding to unprecedented times. indicator of an organization’s performance. cybersecurity and blended experiences may require
organizations to bring along customers who aren’t
ready for the change. Accordingly, the CRO’s role in
creating trust among consumers becomes more
important than ever.
1 2 3 4 5 6 7 8 9 Bespoke for billions: Digital meets physical 147
1
journeys? Are they seamless and Learn how understanding consumers’
consistent across all channels, needs can help brands build emotional
both physical and digital? What bonds at an enterprise scale.
does it feel like from your
customers’ perspective?
2
with technologies that will lead of uncertainty
to offering more personalized Tune in to live, weekly conversations
experiences, both in-person and and insights on how brands can
digital? respond to uncertainty.
3
in a way that is machine-readable?
Explore the 2021 Global Marketing
Can you access customer data
Trends to see how organizations are
beyond demographic and
fostering human connection to meet
transactional data to include
behaviors, attitudes, emotions, the needs of their stakeholders.
and preferences? How can you use
technology to build customer trust?
1 2 3 4 5 6 7 8 9 Bespoke for billions: Digital meets physical 148
AUTHORS
Our insights can help you take advantage of emerging trends. SENIOR CON T R IBU TOR S
If you’re looking for fresh ideas to address your challenges,
let’s talk. Alex Kelleher
Managing director,
Erik Duffield Deloitte Consulting LLP
General manager, Deloitte Digital
experience management platform Rich Whalen
Managing director,
Deloitte Consulting LLP
Deloitte Consulting LLP
eduffield@deloitte.com
Amelia Dunlop
Deloitte Digital chief experience officer
Deloitte Consulting LLP
amdunlop@deloitte.com
Nelson Kunkel
Deloitte Digital chief design officer
Deloitte Consulting LLP
nkunkel@deloitte.com
1 2 3 4 5 6 7 8 9 Bespoke for billions: Digital meets physical 149
ENDNOTES
1. David Segal, “To survive the pandemic, Savile 9. Deloitte Insights, Digital reality: The focus shifts
Row cuts a bespoke strategy,” New York Times, from technology to opportunity—Tech Trends 2018,
November 15, 2020. December 5, 2017.
TREND 9
E
ven before social justice protests drew all people within an organization. Definitions More than half of employees across all
the world’s attention in the summer of of diversity, equity, and inclusion vary based markets said that DEI performance is
2020, business leaders understood the on an organization’s focus and culture, but “extremely” or “very” important when
case for diversity, equity, and inclusion (DEI). generally, a diverse workforce includes a wide deciding where to apply and whether to
But those events spurred many organizations mix of demographics, skills, and experiences, accept an offer. This percentage is even
to elevate DEI as a business imperative. In fact, while equity provides all workers with fair higher for employees in China (89%),
96% of CEOs now consider DEI to be a strategic access, opportunity, resources, and power to Brazil (88%), and the United States (75%).3
priority.1 And striving for social justice is hardly thrive. An inclusive culture supports fairness,
the only reason: An earlier study showed that equity, and respect, giving workers a sense of Though most organizations view DEI as
organizations with inclusive cultures are twice belonging and being valued and empowered. important, there’s a lot of work ahead. In
as likely to meet or exceed financial targets Deloitte’s external 2020 Human Capital Trends
as those without, three times as likely to be Diversity, equity, and inclusion are survey, 79% of participants said fostering
high-performing, six times more likely to be important to current as well as potential a sense of belonging in the workforce was
innovative and agile, and eight times more likely employees worldwide. A 2020 Lenovo/Intel important to their organization’s success,
to achieve better business outcomes.2 global employee study suggests that an and 93% agreed it drives organizational
organization’s DEI policies and performance performance. Unfortunately, only 13% of
Workforce DEI programs and policies encourage are significant factors in a candidate’s decision respondents said their organization is “very
and support representation and participation of to apply for or accept a job. ready” to address the trend.4
1 2 3 4 5 6 7 8 9 DEI tech: Tools for equity 152
We expect enterprises looking to close this play a critical role as strategic partners by These tools could be seamlessly integrated
readiness gap to increasingly use technologies designing, developing, and executing tech- into the work and collaboration software
to support DEI outcomes across all aspects of enabled solutions to address increasingly that workers and managers use every day,
the employee journey—from talent sourcing complex challenges. They can help identify allowing workers to opt in to technology
and selection through employee experience, areas that lack diversity or equity by solutions that nudge behaviors toward
compensation, retention, and development. reengineering the way data is collected, equity in moments that matter. For example,
managed, analyzed, and reported. Technology a manager who is preparing to conduct
Tools can help identify, recruit, • Identify and address biased language in job postings using natural language processing (NLP)
Recruitment and develop, and advance a more • Nudge recruiters at key points in the hiring process to increase awareness of potential bias using AI
diverse talent pool. • Access pools of qualified, diverse candidates through candidate search platforms
advancement • Objectively identify “optimal” candidates for jobs or promotions using AI, machine learning, and automation5
Tools can help leaders build • Support efforts to build inclusion and belonging within work groups using organizational network analysis and community-
Leadership inclusive cultures, including building social platforms
engagement and retention of • Encourage more objective performance reviews using NLP and ML
and culture diverse talent. • Gain insight into behavior changes needed to develop inclusive leaders using behavioral assessment tools and learning platforms
Data and analytics tools can be • Monitor DEI KPIs, including compensation and advancement equity, using advanced analytics, data visualization, and
Measurement used to establish organizational interactive dashboards
baselines, measure progress, and • Match people to diverse workplace opportunities and coaches using data insights
and insights
deliver actionable insights. • Predict which workers are likely to leave using predictive forecasting models to proactively intervene
• Evaluate qualitative and quantitative outcomes of DEI efforts using advanced analytics
1 2 3 4 5 6 7 8 9 DEI tech: Tools for equity 154
Kavitha Prabhakar
workforce DEI as top priorities for the firm and to follow through by
taking actions that accelerate our progress.
Chief DEI officer, The sponsorship and commitment of Deloitte’s board and executive
Deloitte Consulting LLP committee—specifically US board chair Janet Foutty and US CEO
Joe Ucuzoglu—provided the support we needed to make significant
progress quickly. We challenged ourselves to balance the risks that
1 2 3 4 5 6 7 8 9 DEI tech: Tools for equity 157
could result from unconscious inequities generate insights to help create a more diverse, Deloitte was on the verge of implementing
that could be uncovered through talent data equitable, and inclusive workplace. videoconferencing when the pandemic began.
transparency with the risks of not being The visual nature of video calls helps people
transparent, which could threaten our people’s We’re also collecting and securing more self- see the diversity in their teams and encourages
engagement and productivity. reported information about our people’s diversity, inclusion and equitable opportunities for
both visible and invisible. Individuals now have the participation. Of course, there’s room for
Beyond executive sponsorship, an effective option to self-identify their background and other improvement. For example, our IT function has
DEI strategy requires collaboration across the attributes, as well as to opt in to communications taken on the challenge to improve our people’s
organization. Deloitte’s IT function is partnering from affinity groups and other networks that can online visual images. We’re working on an app that
with the businesses, the talent organization, help them engage and network across the firm. will automatically adjust web cameras and lighting
and the DEI team to appropriately incorporate to accurately capture a more diverse range of skin
data and technology tools as one component of We plan to move from using data for hindsight and hair tones and help everyone look their best.
Deloitte’s holistic workforce strategy. insights to predictive analytics that support
decision-making. For example, we would like During recent months, Deloitte has made
One of our goals is to leverage technology to to know the key pieces of historical data that significant progress in collecting and
better understand our workforce and make indicate that a person is likely to leave the firm. By safeguarding appropriate workforce data and
more-informed decisions. To do this, we needed monitoring this data for our current workforce, we analyzing it for meaningful, actionable insights
access to talent data beyond what’s required could aim to identify those who may be flight risks across the talent life cycle. We’re also adopting
for regulatory reporting. Historically, we’ve had so that we have an option to take appropriate and refining technology tools to support a more
a wealth of data available in different systems, preemptive actions. inclusive culture. There’s much work ahead, but
but it was not integrated in a way that was easy we’re dedicated to helping build a more diverse,
to access. So, Deloitte is implementing a people New collaboration tools can also help to build equitable, and inclusive society for the benefit
analytics program to consolidate talent data and inclusion in a virtual environment. Fortunately, of all.
1 2 3 4 5 6 7 8 9 DEI tech: Tools for equity 158
EXECUTIVE PERSPECTIVES
STRATEGY // A growing number of FINANCE // More than ever, leaders RISK // In the arena of social issues,
CEOs have set ambitious goals for DEI are answering to the public for the many often see technology as being
but have struggled to measure progress. organization’s vision of social responsibility. biased. However, the fundamental issue
Emerging DEI tools can help CEOs make the cultural Accordingly, many forward-thinking CFOs are taking with workplace bias lies not with software tools but
shifts needed to reduce bias and create inclusive a stronger role in defining what that vision is, since with the biased humans who build or use them.
workplaces. For this to be successful, CEOs should talent remains a top-five priority for them. With new Tools such as those discussed in this chapter can
fully understand DEI tech capabilities and require workforce entrants aware of employers’ commitment provide the nudges or insights needed to reduce
CIOs and diversity leaders to design purposeful to DEI, social responsibility may soon play a stronger human bias. They can also manage the data that
constraints so that tools are used for good. In some role in retaining talent. CFOs may be less concerned can help hold an organization accountable to its
cases, organizations with advanced DEI analytics have than others with the granular details of inclusion vision of diversity, equity, and inclusion. When
struggled to act on their new insights and created technologies used to monitor recruiting bias, but they chief risk officers (CROs) manage the inherent
more complex problems. With this in mind, CEOs should keep a keen eye on tools that can help train risks of transparency, data gathering and analysis
should work with stakeholders to develop tailored and develop a diverse workforce with needed skill from such tools may help organizations measure
strategies and informed teams for each inclusion sets. Especially in the current virtual environment, progress on DEI. Risk leaders can then look for ways
issue they’d like to address. CFOs can also capitalize on remote work models to to use predictive risk analytics to intervene before
identify and recruit diverse candidates who would issues arise.
normally be unavailable.
1 2 3 4 5 6 7 8 9 DEI tech: Tools for equity 159
1
Which tools and platforms are you in Tech collection
evaluating to support your overall Explore how organizations are pursuing
organizational DEI efforts? and achieving DEI goals within their
technology organizations.
2
efforts focus on spanning the Read how tech leaders can adopt
employee life cycle rather than and integrate new tools into
standalone initiatives such as employee well-being strategies.
reducing recruitment bias?
HR technology’s role in
3
How are DEI and tech leaders activating a diversity and
collaborating to implement inclusion strategy
technology that will provide Learn how HR technology can
meaningful DEI outcomes? provide the data and tools to
improve workforce inclusion.
1 2 3 4 5 6 7 8 9 DEI tech: Tools for equity 160
AUTHORS
Our insights can help you take advantage of emerging trends. SENIOR CON T R IBU TOR S
If you’re looking for fresh ideas to address your challenges,
let’s talk. Bill Docherty Donna Cobb
Managing director, Senior manager,
Christina Brodzik Deloitte Consulting LLP Deloitte Consulting LLP
Principal, Consulting DEI practice lead
Louise Nickson Devon Dickau
Deloitte Consulting LLP
Director, Senior manager,
cbrodzik@deloitte.com
Deloitte MCS Limited Deloitte Consulting LLP
ENDNOTES
1. Deloitte, Insights on CEO priorities and predictions,
October 2020.
Acknowledgments
Executive editors
Scott Buchholz In his role as CTO for Deloitte Mike Bechtel Prior to joining Deloitte, Bechtel led
Emerging technology research Consulting LLP’s Government and Managing director and chief futurist Ringleader Ventures, an early-stage
director and Government & Public Public Services practice, Buchholz Deloitte Consulting LLP venture capital firm he cofounded
Services chief technology officer works with clients to implement mibechtel@deloitte.com in 2013. Before Ringleader, he
Deloitte Consulting LLP innovation across a diverse set served as CTO of the Ounce of
sbuchholz@deloitte.com of areas, including emerging As chief futurist with Deloitte Prevention Fund, a national not-for-
technologies, legacy modernization, Consulting LLP, Mike Bechtel helps profit focused on early childhood
With more than 25 years of experience and solution architecture. As the clients develop strategies to thrive education for at-risk youth. Bechtel
in technology innovation and firm’s emerging technologies research in the face of discontinuity and began his career in technology R&D
implementation, Scott Buchholz focuses director and the sponsor of Tech disruption. He researches the novel at a global professional services
on helping clients transform the way Trends, he helps identify, research, and exponential technologies likely firm where his dozen US patents
their organizations deliver their missions and champion the technology trends to materially impact the future of helped result in him being named
and businesses through technology. He that are expected to have significant business, and builds relationships that firm’s global innovation director.
supports organizations across industries impact on the market and clients’ with the startups, incumbents, and He currently serves as professor
by providing advice and insights on how businesses in the future. Buchholz academic institutions creating them. of corporate innovation at the
to evolve their technology and their also leads Deloitte’s efforts to University of Notre Dame.
organizations to improve performance, explore quantum computing and
effectiveness, and efficiency. quantum technologies.
1 2 3 4 5 6 7 8 9 Acknowledgments 163
S T R AT EG Y F IN A NCE
Ashok Divakaran
Principal | Deloitte Consulting LLP R ISK
Special thanks
Stefanie Heng for being our stupendous, stern Dana Kublin for being our gifted graphical guide, Cheylin Parker, Tracey Parry, Daniella Ramirez,
sherpa throughout the research, development, and taking our fantastic flights of fancy and turning them and Tiffany Stronsky for our marvelous marketing,
publication process, eternally calm under pressure. into intuitive, insightful infographics. Thanks for being continuous communications, and provocative PR.
Your spreadsheets, to-do lists, and nonstop emails our tireless translator between words and pictures, Your ongoing encouragement makes sure that we
kept all of us on track and on our toes, and your overseeing the graphics and artwork that make our continue to take our buzz up to 11.
creative solutions saved us from our own chaos trends better.
monkeys. Matthew Budman, Blythe Hurley, Hannah Rapp,
Caroline Brown, Tristen Click, and Linda Holland and the entire Deloitte Insights team. We appreciate
Anh Nguyen Phillips for being our relentless for being our consummate creative team. Your our amazing and ongoing partnership that helps Tech
research leader, overseeing our overall research collective ability to translate jargon into English and Trends reach new heights every year.
efforts, and deftly laying down the law. Your ability to transform chatter into infographics is second to none.
always be reasonable, be looking out over the horizon, Thanks for always dotting the I’s and crossing the T’s Jodi Gray, Matt Lennert, Mackenzie Odom, Joanie
and be asking good questions helped us deliver under while going the extra mile. Pearson, Samantha Trunzo, Alexis Werbeck, and
pressure with grace. the Green Dot Agency. We appreciate your ongoing
Matt Calcagno, Kelly Gaertner, Natalie Martella, partnership that helps us get the word out and makes
Doug McWhirter for being our wise, witty muse, Abhijith Ravinutala, and Maria Wright for keeping Tech Trends look fabulous.
continuously distilling structure from boatloads of the Tech Trends train running on time. We benefited
brainstorms and brainwaves while culling through from your help keeping us on track across interviews,
innumerable interviews, reams of research, and secondary research, writing, content reviews, graphics,
stampedes of SMEs. We appreciate the poise and dry and so much more.
humor that you bring to our research projects.
Sign up for Deloitte Insights updates at www.deloitte.com/insights. www.deloitte.com/us/TechTrends
About Deloitte
Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee (“DTTL”), its network of member firms, and their related entities. DTTL and each of its member
firms are legally separate and independent entities. DTTL (also referred to as “Deloitte Global”) does not provide services to clients. In the United States, Deloitte refers to one or more of the US member
firms of DTTL, their related entities that operate using the “Deloitte” name in the United States and their respective affiliates. Certain services may not be available to attest clients under the rules and
regulations of public accounting. Please see www.deloitte.com/about to learn more about our global network of member firms.