Business Intelligence Best Practices for Dashboard Design
Keys to Effective Dashboards: Data, Data, Data
A White Paper
by Dan Carotenuto
Dan Carotenuto is a technical director in Corporate Marketing at Information Builders. He is a driving force behind Information Builders’ business intelligence and integration technical marketing materials. His team creates product demonstrations, educational materials, and Flash-based marketing solutions; conducts public relations, press, and analyst product reviews; and researches new and emerging business intelligence technologies. Dan has been with Information Builders since 1989. He often speaks about business intelligence and integration at industry conferences and seminars.
Table of Contents
Overview Visual Design
A “Few”Words on Visual Design Visual Design in Action Word-Sized Graphics
2 2 3 3
5 5 7 11
Dashboard Data Integration
Show Me the Data Integration Styles Integration and Information Builders
The real estate axiom “location, location, location” makes it apparent that the most important attribute for a piece of property is where it is located. For dashboards, think, “data, data, data.” An oftenoverlooked aspect, data is one of the most important things to consider in designing dashboards. Even if a dashboard’s appearance looks professional, is aesthetically pleasing, and includes graphs and tables created according to accepted visual design standards, other issues come into play when assessing the true success of the application. Remember, appearances can be deceiving. It is also important to ask yourself: Is the data reliable? Is it timely? Is any data missing? Is it consistent across all dashboards? Although visual design is important, sometimes the biggest challenge is getting the right data into the right dashboard in the most efficient way. This paper offers an overview of best practice business intelligence (BI) dashboard design principles and discusses data integration options for getting data into a dashboard.
First, let’s make sure we are using the same language with regards to dashboards. In 2004 Stephen Few, a data visualization expert, wrote an article for Intelligent Enterprise magazine that defined a dashboard as “a visual display of the most important information needed to achieve one or more objectives consolidated and arranged on a single screen so the information can be monitored at a glance.”1 In 2007 Gartner expanded the definition to: “…a reporting mechanism that aggregates and displays metrics and key performance indicators (KPIs), enabling them to be examined at a glance before further exploration via additional BI tools. Dashboards are useful KPI and metricreporting mechanisms that enable users to quickly monitor and track performance via an aesthetic user interface. They employ visualization components, such as gauges, thermometers, dials, and traffic lights.”2 From these definitions we can make several agreed upon assumptions about dashboards: A user should be able to look at a dashboard and quickly make observations without scrolling, drilling, or clicking off the initial screen. Minimal user interaction can be included to enhance understanding and clarify observations, but too much interaction defeats the purpose of a dashboard and crosses over into the realm of analysis. And that dashboards also:
I I I
Provide a way to monitor and track performance Should be able to convey what is going on rather quickly Typically contain key performance indicators and use several types of data visualization
A “Few” Words on Visual Design
Since dashboards serve as a way to monitor performance at a glance, graphs, icons, and tabular reports should not be put together in an uncoordinated, unplanned manner. Dashboards must be built methodically, strategically, and with attention to detail. The size, color, and style of a font – such as whether it’s bold or italics – matter more in a dashboard than anywhere else in a business intelligence solution. “Two of the greatest challenges in dashboard design,” says Stephen Few in his book, Information Dashboard Design,“ are to make the most important data stand out from the rest, and to arrange what is often a great deal of disparate information in a way that makes sense, gives it meaning, and supports its efficient perception. An understanding of the preattentive attributes of visual perception and the Gestalt principles provides a useful conceptual foundation for facing these challenges.”3 Achieving at-a-glance observations means making data pop. Designers must manipulate the graphs and tabular reports common to dashboards so the data reflects problems or opportunities – depending on which is important to the user – and stands apart from the rest of the information. This can be done through the use of icons, colors (hues), shapes, and sizes of objects and properties in a graph or tabular report.4
1 Few, Stephen. “Dashboard Confusion.” Intelligent Enterprise, March 20, 2004. 2 Gartner. “Q&A: Important Integration Considerations for Scorecards, Dashboards and Portals.” July 9, 2007. 3 Few, Stephen, Information Dashboard Design: The Effective Visual Communication of Data. O’Reilly Media, Inc.
4 Few, Stephen. IBID
Business Intelligence Best Practices for Dashboard Design
Visual Design in Action
Graphs, tabular reports, and text can be manipulated to make important information stand out. Figure 1 shows a dashboard created with Information Builders WebFOCUS business intelligence platform. Note that all the text is gray except for the problems, which are black and slightly larger. Even the axes values and borders are displayed in gray, which makes the text in black more apparent. Icons further draw the user’s eye to the data that needs attention so that observations can be made.
Figure 1: Sample dashboard created with WebFOCUS from Information Builders that makes use of visual perception and design standards and principles.
A common problem in dashboard design is how to deal with the limited amount of real estate. Among others, two solutions that Few advocates are sparklines and bullet graphs.5 These are graphs that can be displayed in an area no larger than a word. Edward Tufte, the creator of the sparkline describes them as “…small, high-resolution graphics usually embedded in a full context of words, numbers, images. Sparklines are data words: data-intense, design-simple, word-size graphics.”6
Figure 2: An example of a sparkline graph.
Bullet graphs are Few’s “answer to the problems exhibited by most of the gauges and meters that have become synonymous with dashboards.” Radial gauges waste a great deal of space. This problem is magnified when you have many displayed in a single dashboard. Few describes bullet
5 Few, Stephen. IBID. 6 Tufte, Edward R. Beautiful Evidence, Graphics Press LLC. January 2006.
graphs as “…designed to display a key measure, along with a comparative measure and qualitative ranges to instantly declare if the measure is good, bad, or in some other state.”7 Bullet graphs reinforce the notion that a picture is worth a thousand words. They help the viewer quickly understand comparisons made to targets via a graphical representation.
Figure 3: An example of a bullet graph.
Sparklines and bullet graphs both convey considerable meaning on their own, but when combined along with text-based data in a tabular report, they can convey much more information.
Figure 4: This mobile dashboard created with WebFOCUS from Information Builders shows how sparklines, bullet graphs, and other data elements tell more about the data when used together than when used on their own.
WebFOCUS generates these word-sized graphics as well as other display mediums ideal for dashboards. It provides the ability to manipulate the elements of graphs and tabular reports in accordance to dashboard visual design standards and principles. By their very nature, word-sized graphics demand serious data integration considerations. Sparklines are typically rendered from detailed data and can have thousands of data points, while bullet graphs typically use aggregated data. Generating a tabular report containing these graphs can be difficult, and some common challenges include:
Coordinating the detail-data display containing potentially thousands of data values represented by sparklines and the aggregate data display containing a handful of values represented by bullet graphs into a single row Incorporating these graphs and reports along with others so that they can display data from many different areas of the organization all on a single screen
The way to address these challenges is through data integration.
7 Few, Stephen. IBID.
Business Intelligence Best Practices for Dashboard Design
Dashboard Data Integration
The challenges of designing dashboards does not end with knowing which hues to use and how to make information pop. The data in dashboards is used as the basis for making critical business decisions. Designers must consider what data to use and how to make it available and integrate it into a dashboard solution. The answers to the following questions help determine which data integration style should be implemented:
I I I I I
What is the quality of the data targeted? How accurate is it? Should it be real time? Where is it coming from? Does information from systems outside the enterprise need to be included?
Before reviewing the integration styles that can address these concerns, some common elements of data integration – including data access, data quality, and consistency – must be addressed.
Show Me the Data
The following aspects of data integration are paramount to the effectiveness of dashboards and are common to all styles of integration:
I I I I I I
Data access Data quality and consistency Data consolidation Data latency Impact on operational systems Implementation time and cost
Data Access Data access is the most fundamental property of dashboard data integration. It is not just about getting access to data wherever it is; but also involves accessing the data in the most efficient way possible without sacrificing data quality and consistency. Also along for the ride is the ability to join the data across disparate sources and platforms. Software systems such as Web services have opened new doors for data access by allowing applications to tap into previously unavailable information systems. Business intelligence applications can be made available via Web services, effectively turning them into data sources that can be combined, joined, and reconciled with other enterprise data. Data Quality and Consistency The accuracy of information in dashboards affects business decisions. The Data Warehouse Institute (TDWI) reports that, “…data quality problems cost U.S. businesses more than $600 billion a year. Yet, most executives are oblivious to the data quality lacerations that are slowly bleeding their companies to death. More injurious than the unnecessary printing, postage, and staffing costs is the slow
but steady erosion of an organization’s credibility among customers and suppliers, as well as its inability to make sound decisions based on accurate information.”8 Ensuring data quality and consistency provides a single version of the truth throughout the enterprise and needs to be seriously considered when designing dashboards.
Data Consolidation Data consolidation involves combining data from multiple sources into a central system such as a data warehouse. Consolidating data includes many challenges, such as where the data resides, which data to combine, how often it should be combined, and whether or not it should be transformed. Since dashboards often have an amalgamation of information from across the enterprise, data consolidation is fundamental to making data available for a dashboard. Data Latency The dashboard’s primary purpose determines how current its data must be. For example, a strategic dashboard used to help understand long-term business plans may not be concerned with the dayto-day transactions. Data for this dashboard can be current up to the week, month, quarter, or year. Operational dashboards, on the other hand, require real-time or near real-time data to make immediate decisions and solve problems or address opportunities as they arise. It is important that these dashboards experience little or no data latency. Impact on Operational Systems Business intelligence solutions strive to minimize their impact on operational systems. Dashboards used for strategic purposes typically do not require access to operational systems. Though it may seem that operational dashboards absolutely need access to operational systems, it is not always required.
Operational dashboards can be fed data from a data warehouse instead of an operational system. The process is a variation of the traditional data warehouse style of integration known as the realtime data warehouse. There can also be hybrid approaches where parts of the dashboard use a real-time data warehouse and others pull data directly from the operational systems. Which approach is best depends on the dashboard requirements. The bottom line is that dashboards – as part of a BI solution – should minimize their impact on operational systems.
Implementation Time and Cost There are many ways to make data available to a dashboard. Some are faster than others, and costs vary. Which approach is best is driven by business demands, but should not be determined at the expense of data quality and consistency because it can lead to bad decisions. An ideal approach to data integration might be compromised by making dashboards functional sooner rather than later.
Keep in mind that different integration styles demand different implementation times, which impact how fast a dashboard solution can be made available. Operational data access requires considerably less time to implement than a traditional data warehousing style. The trade off is impact on operational systems.
8 Eckerson, Wayne W. “Data Quality and the Bottom Line: Achieving Business Success Through a Commitment to High
Quality Data.” TDWI Report Series. 101 Communication, LLC. January 2002.
Business Intelligence Best Practices for Dashboard Design
The cost to integrate data for use in a dashboard depends on whether the integration solution is built or bought. This decision is determined by the level of customization required and the longterm volatility of the IT infrastructure. Buying a custom solution might mean having to change the business to work with the solution. Building a solution might be necessary when a custom solution simply does not exist.
There are many ways to address the fundamental challenges of data integration. They all revolve around building the right architecture and selecting an appropriate integration style. For dashboards this means choosing one or more integration styles in order to satisfy the needs of users and ensure that the integration solution can change and adapt as the business changes. iWay Software technology from Information Builders along with WebFOCUS supports many integration styles:
I I I I I I
Data warehousing Real-time data warehousing Operational data access Enterprise information integration (EII) Web services Process-driven business intelligence
Data Warehousing Data warehousing involves extracting data from operational systems, transforming it into a universal data model, and making it available to applications such as business intelligence dashboards.
Dashboards used for long-term strategic direction commonly use data that does not change frequently – at most on a weekly basis. Since data latency is not a concern in these cases, data warehousing is a common option. Data warehousing reduces the unnecessary load put on operational systems, especially when the content of a dashboard demands data from multiple sources on multiple platforms. Dashboards commonly include calculations on or transformations of the originating data. A data warehouse reduces the redundancy of data consolidation processing that Data Warehouse would tax an operational system.
Bulk Load Overnight Combine data from multiple systems into a data warehouse
Figure 5: A data warehouse reduces the load dashboards have on operational systems by extracting operational data, transforming it, and loading it into a separate area dedicated for other applications like business intelligence dashboards.
Real-Time Data Warehouse Operational dashboards frequently require real-time or near real-time data. In these cases, data latency is a primary concern. One of the benefits of a real-time data warehousing approach to dashboard data integration is zero data latency.
The key is feeding the data warehouse in real time or near-real time. Transactions and events are captured as they occur, cleansed on the fly, and loaded into the data warehouse. The obvious difference here from traditional data warehousing is that the extraction step occurs as the transactions come in. This further reduces the impact on operational systems, by minimizing the frequency and volume of scheduled extract, transform, and load (ETL) jobs that load and refresh the data for a traditional data warehouse. Even though real-time data warehousing is conceptually complex Data Warehouse Realtime it still uses straightforward integration technologies like enterprise service buses, real-time transformations, adapters, and transaction processes.
Real-time Data Warehouse
Data warehouse updated a transaction at a time
Figure 6: Real-time data warehousing addresses data latency concerns in dashboards by trickle feeding the data warehouse with operational transactions as they come into the enterprise.
Operational Data Access Business intelligence users often want a single view of enterprise data. This requirement sometimes overlaps into the world of dashboards where users want to monitor what pertains to their specific responsibilities. Operational data access can provide dashboard users with a single view of the data, but it greatly impacts the operational systems. For that reason, some take a conservative position and recommend more traditional integration styles such as data warehousing and real-time data warehousing. However, access to detail data from within a dashboard helps users make better decisions, which is where operational data access can help.
Since dashboards cannot always provide all the information users require, it is useful to provide a method for getting more details. Rolling over a bar in a graph or drilling down on an element in a tabular report can bring up context-specific information that enhances dashboard-spawned observations. This analysis event can tap into operational data when necessary, further reinforcing the quality and accuracy of the observation. As long as this on-demand pulling of operational data improves decision-making, it will outweigh the negative impact on operational systems.
Business Intelligence Best Practices for Dashboard Design
Figure 7: Dashboard designers can help enhance the understanding of observations by including drill-downs or rollovers that pull data from operational systems.
Enterprise Information Integration In a perfect world a dashboard uses a single enterprise data warehouse that has all the information it needs. In reality, this is rarely the case. A dashboard often uses multiple data warehouses to get the data it needs. How are these other sources tied together for use in a dashboard? What happens when users need additional or new content from operational systems? What is done while IT works on integrating that data into the data warehouse? Should dashboard users wait until IT catches up?
EII helps solve these dilemmas by providing the ability to link multiple data warehouses and operational systems. EII helps minimize the delay in getting the right information into the hands of decision-makers. This is an important integration solution for the timely delivery of dashboards as the demands of business change. EII is particularly valuable when companies go through mergers and acquisitions, helping multiple systems from different companies rapidly integrate and get EII information to management.
Combine data from multiple systems at report run time
Figure 8: EII lets designers get data into a dashboard faster by linking multiple data warehouses and operational systems.
Web Services Integrating data from the enterprise with applications hosted outside the organization is possible with Web services used as an information resource. For example, you can reconcile data from a data warehouse in your enterprise with a Web service that is available via an application hosted by a
third-party vendor as if they were a single data structure. This extends the possibilities of what can be delivered in a dashboard. Another way Web services make more information available to dashboards is by exposing the intelligence built into existing reporting solutions as reusable Web services. For example, let’s say a report exists that produces the lifetime value of customers. That information might be useful in a dashboard, but how can it be displayed without bringing along the entire report that produces it? Through Web services, this type of information can be exposed as data and integrated into the Web Services data warehouse or used directly within a dashboard.
Web Service Web Service Web Service
Access appearing to be from a relational table
Figure 9: Web services provide a way to get information from outside the enterprise into a dashboard. They can also expose existing BI reports as data for integration into a dashboard.
Process-Driven Business Intelligence “Work smarter, not harder” is a common axiom in the business world. It is also what process-driven business intelligence does for an organization through the automaton of decision-making. This automation is accomplished by embedding event-driven business intelligence functions into business processes to reduce the need for a physical action or increase the timeliness of a response. Some key business intelligence elements that can be embedded are data visualization, analytics, alerts, and reports.
Supply-chain management is a common area where business processes can be automated. For example, inventory levels can trigger an event that parts need to be ordered from suppliers. An event can be automated to start reordering parts as part of a process flow. A report can show how many parts are needed and include price quotes and delivery dates acquired from suppliers via Web services. The automated system can also use the quotes to select a supplier by determining which supplier has the best price and then generate an order. Dashboards can be used to monitor, improve, and streamline process-driven business intelligence. In the supply-chain example, there might be room for improvement. A dashboard can be used to show that on-time delivery of parts has improved, but profits are flat. Further analysis might indicate that profits are flat due to manufacturing costs. Though intelligence was added to the automation of the supplier price-evaluation process, additional intelligence could be added to factor in turnover history for a part. Instead of re-ordering popular parts often, what if a higher volume was
10 Business Intelligence Best Practices for Dashboard Design
ordered less often? That might result in a better price from suppliers, which in turn means lower costs and higher profits. The dashboard is where this type of observation is made. Analysis and action resulted in improvements to the process-driven business intelligence solution. Operational dashboards can also provide feedback on bottlenecks in the business process, which can help IT decide where to concentrate efforts when implementing or improving a process-driven business intelligence solution. For example, problem resolution time could be displayed in a dashboard to help operations understand where delays exist, how long they last, and which ones are the most adverse. A more targeted and strategic approach can then be used to improve efficiency. Thus, business processes can drive business intelligence and business intelligence can be Process Integration used to drive the business process.
Electronic Transaction Operational System
Figure 10: Process-driven business intelligence helps organizations work smarter, not harder, by automating decision-making. Dashboards can help monitor, improve, and streamline the processdriven business intelligence solution.
Integration and Information Builders
iWay Software’s technology supports over 280 native interfaces on any platform and empowers dashboard designers to use any data from any source – within the enterprise and beyond. Flexibility and agility are recognized by both iWay and WebFOCUS as integral parts of any data integration strategy. Support for packaged applications as well as the ability to create custom interfaces lets organizations choose the optimum solution for their business needs and IT infrastructure requirements. Organizations can pull data from any source, transform it, and load it into any target on any platform. iWay supports all brokers, message queues, and transaction processing monitors. Its ability to access legacy systems gives organizations the flexibility to use the integration style of their choice, from connecting dashboard users to data warehouses and operational data, to hybrid approaches that combine access to traditional data warehousing solutions as well as the ability to drill through to operational systems. WebFOCUS and iWay enable enterprises to build and manage their business processes end to end. This service-oriented approach to process-driven business intelligence means that organizations can create and deploy business intelligence as part of any process or service. This approach allows each service to be managed independently in a plug-and-play manner, minimizing application maintenance efforts and cost.
11 Information Builders
Dashboards help organizations make better business decisions. The way in which they are created is not trivial and attention must be paid to the details. Employing visual design standards and principles are important, but are only part of the journey. The way in which data is integrated from the enterprise is just as important. It requires proper planning to address what users need to see, where the data is coming from, how soon it can be delivered, and ensuring that it is accurate and consistent. Information Builders’ iWay technology provides the options necessary for integrating enterprise data for use in dashboards. WebFOCUS lets designers build dashboards that take advantage of accepted visual design standards. From its support of over 280 native adapters to its ability to automate business processes, these solutions provides organizations with the luxury of being able to select an integration style that meets the needs of their dashboard users while adapting to their enterprise infrastructure.
12 Business Intelligence Best Practices for Dashboard Design
Sales and Consulting Offices
I I I I I I I I I I I I I I I I
I I I I I I I I I I
Atlanta,* GA (770) 395-9913 Baltimore, MD Consulting: (703) 247-5565 Boston,* MA (781) 224-7660 Channels, (800) 969-4636 Charlotte,* NC Consulting: (704) 494-2680 Chicago,* IL (630) 971-6700 Cincinnati,* OH (513) 891-2338 Cleveland,* OH (216) 520-1333 Dallas,* TX (972) 490-1300 Denver,* CO (303) 770-4440 Detroit,* MI (248) 641-8820 Federal Systems,* DC (703) 276-9006 Hartford, CT (860) 249-7229 Houston,* TX (713) 952-4800 Los Angeles,* CA (310) 615-0735 Metro Branch (NY/NJ),* NY Sales: (212) 736-7928 NJ Sales: (973) 593-0022 Consulting: (212) 736-4433, ext. 4443 Minneapolis,* MN (651) 602-9100 New Jersey* (973) 593-0022 Orlando,* FL (407) 804-8000 Philadelphia,* PA (610) 940-0790 Phoenix, AZ (480) 346-1095 Pittsburgh, PA (412) 494-9699 St. Louis,* MO (636) 519-1411 San Jose,* CA (408) 453-7600 Seattle,* WA (206) 624-9055 Washington,* DC Sales: (703) 276-9006 Consulting: (703) 247-5565
Belgium Information Builders Belgium Brussels 32-2-7430240 France Information Builders France S.A. Paris 33-14-507-6600 Germany Information Builders (Deutschland) Dusseldorf 49-211-523-91-0 Eschborn 49-6196-77576-0 Munich 49-89-35489-0 Stuttgart 49-711-7287288-0 Netherlands Information Builders (Netherlands) B.V. Amsterdam 31-20-4563324 Portugal Information Builders Portugal Lisbon 351-217-217-491 Spain Information Builders Iberica S.A. Barcelona 34-93-344-32-70 Bilbao 34-94-425-72-24 Madrid 34-91-710-22-75 Switzerland Information Builders Switzerland AG Dietlikon 41-44-839-49-49 United Kingdom Information Builders (UK) Ltd. London 44-845-658-8484
Information Builders (Canada) Inc. I Calgary (403) 538-5415 I Ottawa (613) 233-0865 I Montreal* (514) 421-1555 I Toronto* (416) 364-2760 I Vancouver* (604) 688-2499
Information Builders Mexico I Mexico City 52-55-5062-0660
Information Builders Pty. Ltd. I Melbourne 61-3-9631-7900 I Sydney 61-2-8223-0600
Austria Raiffeisen Informatik Consulting GmbH Vienna 43-12-1136-3870 Brazil InfoBuild Brazil Ltda. São Paulo 55-11-3285-1050 China InfoBuild China, Inc. Shanghai 86-21-5080-5432 Rongji Software Technology Co., Ltd. Beijing 86-10-5873-2031 Ethiopia MKTY IT Services Plc Addis Ababa 251-11-5501933 Finland InfoBuild Oy Espoo 358-207-580-843 Greece Applied Science Athens 30-210-699-8225 Guatemala IDS de Centroamerica Guatemala City 502-2361-0506 Gulf States Nesma Advanced Technologies I Bahrain I Kuwait I Oman I Qatar I Yemen I United Arab Emirates Riyadh 96-1-465-6767 India InfoBuild India Chennai 91-44-42177082
Israel NESS A.T. Ltd. Tel Aviv 972-3-5483638 Italy Selesta G C Applications S.P.A. Genova 39-010-64201-224 Milan 39-02-2515181 Torino 39-011-5513-211 Japan K.K. Ashisuto Osaka 81-6-6373-7113 Tokyo 81-3-5276-5863 Malaysia Elite Software Technology Sdn Bhd Kuala Lumpur 60-3-21165682 Norway InfoBuild Norway Oslo 47-23-10-02-80 Philippines Beacon Frontline Solutions, Inc. 63-2-750-1972 Russian Federation FOBOS Plus Co., Ltd. Moscow 7-495-124-0810 Saudi Arabia Nesma Advanced Technology Co. Riyadh 996-1-4656767 Singapore Automatic Identification Technology Ltd. 65-6286-2922 South Africa Fujitsu Services (Pty.) Ltd. Johannesburg 27-11-2335911 South Korea Unitech Infocom Co. Ltd. Seoul 82-2-2026-3100 Sweden InfoBuild AB Kista 46-735-24-34-97 Cybernetics Business Solutions AB Solna 46-7539900 Taiwan Galaxy Software Services Taipei 886-2-2586-7890 Thailand Datapro Computer Systems Co. Ltd. Bangkok 662-679-1927, ext. 200 Venezuela InfoServices Consulting Caracas 58-212-763-1653
Sales, ISV, VAR, and SI Partner Information (800) 969-4636
**Training facilities are located at these branches; additional locations are available. **Authorized to sell iWay Software only.
Corporate Headquarters Canadian Headquarters For International Inquiries
Two Penn Plaza, New York, NY 10121-2898 (212) 736-4433 Fax (212) 967-6406 DN7505699.0907 www.informationbuilders.com firstname.lastname@example.org 150 York St., Suite 1000, Toronto, ON M5H 3S5 (416) 364-2760 Fax (416) 364-6552 +1(212) 736-4433
Printed in the U.S.A. on recycled paper
Copyright © 2007 by Information Builders, Inc. All rights reserved.  All products and product names mentioned in this publication are trademarks or registered trademarks of their respective companies.