Organizational Simplification Framework
Organizational Simplification Framework
net/publication/322364868
CITATION READS
1 2,845
1 author:
Nicolay Worren
Norwegian University of Life Sciences (NMBU)
40 PUBLICATIONS 920 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Nicolay Worren on 10 January 2018.
Nicolay Worren
Norwegian University of Life Sciences (NMBU)
School of Economics and Business
P.O. Box 5003
1432 Aas
Norway
Tel: +47 91 000 693
E-mail: [Link]@[Link]
Abstract
several surveys. The purpose of this paper is thus to outline a theoretically grounded framework
for understanding complexity and for identifying opportunities for simplification, particularly
with regard to organizational structures. The framework integrates three different systems
theoretic perspectives that address different aspects of complexity. The functional perspective
focuses on the relationship between the purpose or function of the organization, and its structural
design. The information processing perspective focuses on the work processes and the flow of
information in relation to the formal structure. Finally, the stratification perspective addresses
the governance structure and vertical reporting relationships in organizations. Specific design
principles are derived from each perspectives. The principles essentially operationalize the
concept of the modular organization by specifying how one can create semi-independent sub-
2
INTRODUCTION
- Stuart Kauffman
several surveys that have included thousands of managers around the world (e.g., Economist
Intelligence Unit (EIU), 2011, 2015; IBM, 2010; KPMG, 2011; PwC, 2006). The general
perception is that the level of internal complexity is increasing. The respondents also believe that
complexity has a negative effect on their organization. They indicate that there are conflicting
priorities, that too much time is spent on coordination, and that they are unable to make timely
decisions. The end result is decreased organizational performance. Based on a study of 200 large
firms, Collinson & Jay (2011) estimated that these firms loose on average 10% of their EBITDA
each year due to internal complexity. Although there are several sources of complexity,
organizational structure has been singled out as a key factor. In the 2015 EIU survey, 55% of the
respondents indicated that the organizational structure in their firms were very or extremely
Given these results, it is not surprising that firms are attempting to simplify. In one study,
no less than 97% of the respondents indicated that they were engaged in at least one
books have been published that offer advice to managers, including Ashkenas (2010), Collinson
& Jay (2012) and Maeda (2006). Consulting firms such as BCG and Bain have also seen this
3
need and have developed specific offerings related to complexity reduction that promise their
clients a simpler, flatter, and more effective organization (Morieux & Tollman, 2014; Torsten
Unfortunately, there is little academic research on this topic and we lack systematic
knowledge about the effects of simplification efforts. But the indications that we have from
surveys with managers suggest that these efforts have, at best, a modest success rate. As an
example, in the 2011 KPMG study, 90% of the 1400 senior executives who participated said that
their organization’s success was dependent upon managing complexity, yet less than half
believed that the actions they were taking to manage complexity had been very effective (the
2015 EIU survey came to the same conclusion) (Economist Intelligence Unit, 2015; KPMG,
2011). This may be contrasted with the results from simplification efforts in other areas, such as
in product design and manufacturing, where studies show that modularization - a key means of
managing complexity - has generally had a positive effects on costs, flexibility and overall
financial performance (e.g., Jacobs, Vickery, & Droge, 2007; Piran et al., 2017; Worren, Moore,
Numerous suggestions have been offered for managers who seek to simplify their
organization. A common theme is that of “variety reduction”. The underlying assumption seems
to be that complexity is related to excessive variety or differentiation, and that one can simplify
by reducing variety, for example, by discontinuing some products or minimizing the number of
steps in a work process. Related more specifically to organizations, some authors focus on
complex roles or management processes. On the [Link] blog, Ashkenas & Bodell (2013)
4
“An easy starting point for simplification is to get rid of stupid rules and low-value
activities, time-wasters that exist in abundance in most organizations. Look, for example,
at how many people need to review and sign off on expense reports or small purchases; or
how many times slide decks need to be reviewed before they are presented”.
Cross & Gray (2013) conducted a detailed social network analysis in a firm where they identified
how people requested and provided information to each other. They found that 150 people
indicated that frequently sought out one specific manager for information, and that another 77
said that they needed even more of this manager’s time to meet their business goals. They
concluded that managers and internal experts can easily become overloaded in modern
organizations. As help to individuals in the same situation, they suggest that one should “not be
too responsive or quick to help out with issues” and the one might try to “co-create solutions
with employees so that they take ownership and need fewer interactions with you over time” (p.
10). A third area that is frequently mentioned is the reporting structure or the vertical
management layers within the firm. As stated in Couto, Caglar, & Plansky (2016) “unless
periodically reviewed and adjusted, unwieldy organizational structures burden a company with
high management costs and stultifying bureaucracy” (p. 154). A number of authors have
proposed that the organizational structure should be as flat as possible (Ashkenas, 2010;
All of these suggestions are sensible and may prove useful in some situations. However, a
key limitation is that they mainly deal with effects, rather than underlying causes. If a
management process requires too many approvals, it may indeed be an easy fix in some cases.
But in other cases, this is merely a symptom of a more fundamental issue. When many managers
are required to approve a given action, it may suggest that there are overlapping roles and
5
accountabilities, which in turn may be due to the way the organizational structure has been
heed the advice of Cross & Gray (2013) and remove oneself from interactions in order to reduce
one’s workload. However, the question is whether a re-design, such as a more logical grouping
of roles into teams and sub-units, could have reduced the time and cost of collaboration.
Similarly, reducing the number of management layers may be the right solution for a firm with a
tall hierarchy, but if one flattens the structure too much, it may lead to work overload for
managers, which in turn has a negative effect on performance (Wallin, Pousette, & Dellve, 2014;
Hence the key question addressed in this paper is how we can build a theoretically sound
framework for understanding organizational complexity, and secondly, how we – based on this
framework – can derive design principles that can be basis for interventions aimed at simplifying
organizations.
The examples above suggest that complexity has multiple facets. It also seems clear that
one needs to relate concrete and observable instances of complexity to underlying causes in order
to achieve simplification. For this reason, I have chosen to use systems theory as the starting
point. A key assumption in systems theory (Ackoff & Emery, 1972; Jackson, 2000) is that one
has to view the phenomenon that one is trying to understand as part of a larger whole. The
considering the larger system in which the individual is embedded. In order to improve
performance, the question is thus how we can modify or improve the design of the overall
6
However, there is not one, but multiple “systems theories” (Jackson, 2000). Here I have
selected theories that I group into three “perspectives”. The first is the functional perspective,
which focuses on the relationship between the purpose (or function) of the organization, and its
structural design. The second is the information processing perspective, which focuses on the
work processes and the flow of information in relation to the formal structure. The third is the
stratification perspective, which focuses on the formal structure and vertical reporting
relationships in the organization. The three perspectives are largely independent and
complementary, in the sense that they address different aspects of complexity (Figure 1).
The perspectives that I cover do not represent the entire breadth of theories in the field of
viewed as the main cause of diseconomies of scale. More recently, neo-institutional theorists
stakeholder expectations (Greenwood, Diaz, & Li, 2010; Pache & Santos, 2013). These theories
are already well known and extensively covered in the literature. My purpose here was to select
7
theories that are not necessarily new themselves, but underappreciated, and that may contribute
to an alternative way of framing the issue of organizational complexity. While the predominant
focus in the literature is on descriptive theory, the purpose here is to describe theories that can be
used prescriptively (van Aken, 2004) and provide solutions that can help reduce complexity.
More specifically, this is done by deriving design principles that can be the basis for
discussion to issues related to organizational structure, but similar principles can be used for
In systems theory, one assumes that each organization (as well as each sub-unit within the
functions (Ackoff, 1971; Ackoff & Emery, 1972). A function is fulfilled by a structure (e.g., an
organization or a sub-unit), but the terms function and structure are conceptually distinct (Gresov
& Drazin, 1997). When designing a system, one starts by identifying purposes and functions
(i.e., ends), and then proceeds by considering structures (i.e., means) that may fulfil the
means are formalized as design parameters (DPs) (typically physical components; but they may
also be software components). It is also possible to identify functional requirements for social
systems, including organizations. The functions may be deduced from already existing
documents that describe missions, unit mandates, project charters, and so on, and may be further
clarified in interviews with members of the organizations. Design parameters for an organization
8
are the organizational units, temporary projects or individual roles that are made accountable for
In axiomatic design theory (Suh, 1990, 2001), one considers the relationship between
functional requirements and design parameters in more detail. One assumes that every
organization can be described by decomposing a high level function (e.g., “Provide software
products for the health sector”; “Offer attractive banking services”) into more detailed functions,
which are linked to specific design parameters (Worren, 2018b). For an organization, this means
that one defines a hierarchy of functionality that corresponds to the hierarchy of authority (i.e.,
the formal structure of the organization) (see Figure 2). This constitutes a new manner of
representing organizations as systems. Unlike an organization chart that only shows the reporting
structure, one can link the structural components of an organization (e.g., roles, teams, sub-units)
with their purpose. Similarly, unlike corporate “mission statements”, which are rarely
operationalized and decomposed, one can link higher and lower-level missions or unit mandates
(i.e., functions). The dual focus on both function and structure is a unique feature of the
functional perspective, and makes it possible to re-frame some common organizational design
Figure 2. Mapping between functional requirements (FRs) and design parameters (DPs) of a
hypothetical software firm. The arrows indicate that when designing an organization, one starts by
9
defining a high level FR and then iterates between DPs and FRs until one has a design that can be
implemented.
Suh (1990) introduced a tool, called the design matrix, to map the relationship between
FRs and DPs (Table 1). The FRs (usually formulated as verbs) are listed vertically and the DPs
(usually formulated as nouns) horizontally. Inside the matrix, one indicates which DP that fulfils
which FR. In the case shown in Table 1, this is done by using a capital “X” for the intended
relationship (i.e., the designation of the unit or role that is accountable for achieving a given FR).
One may also quantify the relationship, that is, identify the extent to which a given DP affects a
Design parameters
Product Customer
R&D Sales
Functional requirements development support
Table 1. Design matrix for hypothetical software firm (the same as in Figure 2).
According to Suh (1990) it is axiomatic that one should maintain the independence of the
functional requirements (FRs). This principle implies that one must first define the functional
requirements, and then design the organization (specified as design parameters (DPs)) such that
each of the FRs can be satisfied without negatively affecting the other FRs (i.e. the FRs are
independent). When this is not the case, one has a design with functional conflict (i.e., a conflict
10
or contradiction between functional requirements) (Worren, 2018a), which Suh (1990) called
coupling.
There are multiple causes of functional conflict. One common source is competition for
resources, as when two product divisions share the same manufacturing facility, which is unable
to meet the demand from both of the divisions (if it prioritizes one division, it will negatively
affect the revenues of the other). A second cause is different time orientations, as when a sub-
unit is responsible for two sets of activities, one that requires quick feedback to internal or
external customers (e.g., customer support), and another that requires more time consuming
analysis (e.g., error identification and mitigation); the general tendency is to spend most of the
time and resources on the more urgent tasks. A third possible cause is role conflicts, for example,
a staff unit that develops company policy that also performs internal services and which creates
policies that favour (or is perceived to favour) its own service providing activities.
accountability for teams and sub-units, and creates goal conflicts as the achievement of one sub-
unit’s goals may compromise the ability of another sub-unit to achieve its goals. It increases
uncertainty and causes a status quo bias, as it becomes less clear for decision makers which
levers they should pull to achieve a given outcome (Ethiraj & Levinthal, 2009). Finally, it may
also lead to role conflict and ambiguity at the individual level, for example, when individuals in
matrix roles report to two or more managers who pursue conflicting goals (Rizzo, House, & and
Lirtzman, 1970). According to Suh (1990), functional conflict (or what he termed coupling)
represents complexity and reduces the probability that the organization will be able to achieve all
of its goals. We should note that this definition differs from the more common conception of
11
complex if the FRs can be satisfied independently (i.e., if there is no functional conflict, or
coupling) (Suh, 2001). This means that it may not be an effective strategy to minimize
complexity in a firm by, for example, reducing the number of product variants or closing down
offices: This reduces the “functionality” (the outputs delivered by the organization), whereas the
axiomatic principles are intended to allow the organization to deliver the same outputs, but with
steps. First, one needs to identify the firm-specific functions (functional requirements) of the
many organizations to have many implicit and inconsistent goals (Argyris & Schön, 1978).
Hence one usually needs to help stakeholders clarify and agree on what the actual goals are for
their organization. The second step is to map the functions or goals against the design
parameters, that is, determine how they are satisfied by specifying the sub-units, projects, or
roles etc. that contribute to the fulfilment of the functions. The result of this step may be called
the “intended design matrix”, as it will identify the formally accountable units for the different
FRs. The third step is to assess the degree of functional conflict or coupling. As suggested in
Worren (2018a), this may be done by making pairwise comparisons of the FRs against each
other (“When DP1 implements measures to maximize FR1, what effect does this typically have
on the ability of DP2 to maximize FR2?”). Based on this exercise, one will be able to produce a
design matrix that shows the actual correspondence between FRs and DPs, including unintended,
negative relationships.
The fourth step is to consider how one can remove the conflicts and contradictions. One
can distinguish between two main approaches. One may first try to work within the constraints of
12
the existing organizational structure. The most common solution is structural separation. This
involves splitting an existing sub-unit into two parts or the establishment of a new sub-unit,
which is allocated one of the functions. The key assumption is that firm can continue to perform
the same functions, but that they need to be allocated to different internal units so that one avoids
functional conflict at the sub-unit level (for some empirical support, see Ethiraj & Levinthal,
2009 and Huckman & Zinner, 2008). Consider a situation where there are two software
development teams, each working on a new software product. The key goals for the two teams
are to develop the new products according to the specifications - and on time and on budget. In
addition, they have been asked to ensure that the finished products conform with quality
individual programmers, and on the total system (Table 2). However, the team leaders gradually
realize that there is a contradiction between these two goals. It requires resources to perform
testing, and one of the programmers in each team has to set aside his or her core tasks to do the
testing, which slows down the coding work. It is difficult for that person to give critical feedback
to peers who are members of the same team, and asking an internal person to perform this work
also reduces its legitimacy: People elsewhere in the firm suspect that quality is not taken
seriously. In addition, the team leaders realize that different skills and tools may be needed to
perform effective testing, compared to the software development tasks. Hence they decide to
create an independent team, which performs testing for both of the development teams (Table 3).
The test team will work in parallel with the development teams, meaning that they will, at the
end of each day, make their working build available to the test team. This solution resolves the
13
Design parameters
Development Development
Functional requirements team 1 team 2
Ensure quality X X
Table 2. Design matrix for software firm with two development teams.
Design parameters
Development Development
Test team
Functional requirements team 1 team 2
Table 3. Design matrix for software firm after establishment of separate test team.
14
Although structural separation is the most common solution, another possible solution is
temporal separation, where the same firm or sub-unit is engaged in different and potentially
conflicting activities, but at different points in time. A medical center may provide medical care
to patients Monday to Thursday, but spend Fridays performing clinical trials or engaging in
research (cf. Huckman & Zinner, 2008). A firm may focus its effort on achieving sales growth
for the five first years, and then prioritize cost reduction the next five years (cf. Ethiraj &
Levinthal, 2009). In some cases, neither structural nor temporal separation is an option, and one
has to introduce a new organization logic in order to remove the functional conflict. One
example may be a product based firm that introduces a “solutions” strategy where it seeks to
combine individual products into a combined offering. This will usually be difficult if the main
units are still measured on product profitability, as the profitability of individual products may be
reduced by creating joint offerings (even if the firm benefits as a whole). This conflict is
removed by restructuring the firm according to market units instead of product units (Day, 2006;
Galbraith, 2002).
Organizations consist of sub-units that not only need to process information to perform
their own tasks, but that also rely on information from other sub-units (or external sources) in
order to plan and execute these tasks. This is the key assumption behind the information
processing perspective, which was established by early writers in the field such as Galbraith
(1974), Daft & Lengel (1984) and Tushman & Nadler (1978). A key variable is the existence of
a dependency (or interdependency) between two or more roles or sub-units. Information must be
transferred or exchanged between units that are interdependent in order to ensure that a work
15
process is completed and that one achieve the desired outcomes (regarding time, cost, quality,
etc). Yet information exchange and coordination may be time consuming and costly activities.
For this reason, one should consider how can minimize coordination costs by finding logical
These basic ideas have been developed further by scholars within engineering
management (Eppinger & Browning, 2012) and modular systems theory (Clark & Baldwin,
2000). An important contribution has been the development of the Design Structure Matrix
(DSM), at tool that can be used to analyze interdependencies between tasks or (alternatively)
between roles and sub-units in an organization. The DSM is a square matrix where
interdependencies between elements (tasks, roles, sub-units etc.) are shown (Figure 3). Reading
across a row, one sees from which other element in the system (task, roles, sub-units etc.) a given
element needs input (e.g., information, resources, approval, etc.) (Unlike the design matrix
described above, which maps functional requirements against design parameters, the DSM thus
maps design parameters against design parameters). The DSM can first be used to document and
represent the current interdependencies in an organization. It can then be used to optimize the
grouping or clustering of elements. The basic assumption is that information exchange and
coordination are more time consuming and costly if they occur across rather than within sub-
units. If there are interdependencies between sub-units, it will require coordination with people
who report to a different manager, who probably pursue other goals and priorities, and who may
also be located in a different office (Kilmann, 1983). Hence a key goal is to align the
interdependencies between tasks or roles related to a work process with the formal structure of
the organization (i.e., the sub-unit boundaries). There now exist algorithms that can take DSM
16
data as input and cluster interdependencies in a manner that minimizes the number of
interdependencies within clusters (e.g., Yu, Goldberg, Yassine, & Chen, 2003).
Figure 3. Example of clustering of DSM elements. The X’s represent interdependencies between
the elements (e.g., element B is dependent upon input from C, but C is not dependent upon input from B).
Comparison to a graph representation of the same data (Worren, 2018a).
organization where the overall coordination load is high (Nuñez, Giachetti, & Boria, 2009) and
where the formal structure is misaligned with the work process interdependencies (Browning,
2001; Kilmann, 1983). This situation not only increases coordination costs but also reduces
flexibility, because changes cannot be made independently in one sub-unit without affecting the
performance of other sub-units. It may also lead to a less delegation of authority and to a need
17
for intervention by managers at higher levels, in order to resolve conflicts and disagreements
between interdependent sub-units (Kilmann, 1983; Stea, Foss, & Foss, 2015).
used for coordination, one may first consider whether one is able to remove one or more
interdependencies. Baldwin & Clark (2000) describe how some decisions by engineers
responsible for different modules of a product, such as a PC, may be reciprocally interdependent,
which may lead to unwanted iterations in the work process (as one engineer requests information
from a second engineer, who cannot provide the information before the first engineer has
completed his or her task). The question is whether it is possible to formulate a “design rule” up
front that can substitute for mutual adjustment and coordination during the work process. One
example is standards for manufacturability, which may substitute for coordination between
design and manufacturing engineers regarding the manufacturability of products that are being
developed.
If one has identified the interdependencies, and ascertained that they cannot or should not
be removed, the next step is to group the tasks or roles to minimize coordination costs (i.e., to
align the formal structure with the work process interdependencies). This is a criterion that has
been described by a number of authors (e.g., Kilmann, 1983; Thompson, 1967; Tushman &
"In partitioning, choose the elements so that they are as independent as possible; i.e.,
elements with low external complexity and high internal complexity" (Alexander,
1964).
"In partitioning a system into subsystems, choose a configuration with minimal
communications between the subsystems."
"Don't partition by slicing through regions where high rates of information exchange
are required."
18
The challenge is that work process interdependencies are rarely documented in the same manner
as the formal organization, and consequently may be ignored or misperceived (Sherman &
Keller, 2011). One therefore needs to interview participants in a work process or distribute an
electronic survey to collect this information, before one can use the DSM to map and optimize
the grouping of roles (for an example, see Worren, Soldal, & Christiansen, 2017).
Even if the intention is to group related roles in the same sub-unit and reduce
coordination across boundaries, one rarely ends up with completely isolated sub-units when
using the DSM procedure. On the contrary, there are usually some “residual interdependencies”
(Mintzberg, 1979) that remain between the sub-units. Once units are formed, one needs to
will be effective will depend on the circumstances. The key variable is uncertainty (cf. Ven,
Delbecq, & Koenig, 1976). With relatively predictable tasks, one may rely on planning and
procedures to coordinate the actions of multiple sub-units. With less predictable tasks, one will
need to move to less formalized and more intensive forms of coordination, such as video
conferencing and meetings. More coordination is generally associated with faster and better
integration (e.g., two engineering teams, each responsible for a component, will be able to
integrate their outputs into one product more easily when the team leaders of the two teams
coordinate effectively with each other). However, more intensive forms of coordination are more
costly, and just as managers may sometimes fail to coordinate effectively, they may also
sometimes select coordination mechanisms and processes that are costlier than they need to be,
19
THE STRATIFICATION PERSPECTIVE
“stratification perspective” are theories concerned with the structuring of the vertical layers of
roles (i.e., the reporting structure in the firm) and the allocation of decision rights and
It is a common assumption that the vertical structure of a firm, and in particular, the
number of management levels, is related to both costs and complexity. New information
technologies have increased the capacity of senior executives and made many middle manager
roles redundant (Wulf, 2012). Consequently, during the 1980’s large firms started to “delayer”,
that is, systematically eliminate layers in the hierarchical structure, particularly at the top of the
organization. In the U.S., the average span of control for CEOs almost doubled from 1985 to
2005 (from 4.7 to 9.8 direct reports on average) (Wulf, 2012). The trend continues. For example,
KLM, the Dutch airline, announced in 2016 that it would introduce a flatter organizational
model. The airline’s CEO, Pieter Elbers, stated: “We are currently going through a process to
make the organization simpler" ([Link]., 2015). The changes are expected to lead to a
reduction of 25% of its managers. DXC, the IT technology firm, announced in August 2017 that
it intended to save US$1 billion by trimming four layers of management (from 11 to 7 layers of
management) (Novinson, 2017). FMC Subsea, an engineering firm, reduced the number of
management levels in its regional units from 8 to 3 as part of a reorganization in 2015 (Worren,
2017). As a result of budget cuts, even public sector organizations in many countries have been
forced to cut management layers to become more cost-effective (Morris & Farrell, 2007).
20
How, exactly, does the number of levels in a firm influence its effectiveness, and more
specifically, the complexity of its organization? A higher number of layers generally leads to
higher costs, not only because of the additional number of roles, but because resources at higher
levels of the organization are more expensive. In a study of 663 Swiss firms, Kuhn (2011) found
that delayering significantly increased profitability, and one would presume that this is partly
because of a reduction of costs. Other studies have shown that delayering is often accompanied
by downsizing, and typically motivated by a desire to reduce costs (e.g., Rajan & Wulf, 2003).
However, there are also many indirect effects of having a steep hierarchy, as stated by
“Every additional level makes the attainment of common direction and mutual
understanding more difficult. Every additional level distorts objectives and misdirects
attention. Every link in the chain sets up additional stress, and creates one more source of
inertia, friction and slack (…)”
One effect is a delay in decision making processes. Eberhardt Rechtin, former president
of the Aerospace Corporation in the US, claimed that the time that it takes to reach a final
decision doubles with each added approval that is required. So, as an example, if the time that it
takes to get the first approval is 30 days, it will take 240 days to reach the final decision if there
are four levels of management that need to get involved (Rechtin, 1991). The number of layers
that a request from the lowest level of the organization would need to pass through before it can
be approved has been called “organizational distance”. Banet, Piehl, & McCann (1976)
interviewing managers and correlated the results with organizational effectiveness. They
concluded that a distance between two and three was optimal for these firms.
21
Another effect is filtering and aggregation of information as it passes through the levels
of a hierarchy (Mintzberg, 1979). In a recent study of lending decision in a large Indian bank,
Skrastins & Vig (2015) showed that lending decisions that are made at higher levels of the
organization are made based on different criteria compared to decisions made a lower levels: At
higher levels (and in larger banks), one relies less on “soft information” about potential
customers and uses more standardized and formalized indicators (e.g., credit history) when
approving loan applications. They also found that lower usage of “soft information” led to worse
performance, and that the return on loans decreased by 14 percent by adding one managerial
layer.
issue, as pointed out by Jaques (1989). If there are more management levels than needed, it
means that the span of control is low. A manager with only a few subordinates will have a
surplus of time that will probably be filled by performing operational rather than managerial
tasks. The roles and responsibilities of the manager versus his or her subordinates are thus likely
to overlap. Conversely, with fewer management levels, the span of control are larger and the
manager will by necessity be forced to give employees more autonomy and focus his or her
efforts on key managerial tasks (e.g., planning and coordinating the work, following up the
However, it is also well documented that delayering can go too far. When the span of
control increases, there is a higher risk of “work overload” for managers, i.e., having too many
responsibilities given the time available. One study that surveyed managers in 14 hospitals found
that a higher span of control was associated with self-reported adverse outcomes (i.e., medical
22
errors, infections and injuries) (Wong et al., 2015). The relationship between the number of levels
(Meier & Bohte, 2003). Initially, a reduction of levels in a steep hierarchy leads to cost reductions
because management positions are removed. It may also increase motivation among employees as
a result of higher autonomy. However, these gains are subject to diminishing marginal returns.
Excessive delayering may reduce performance. It may lead to work overload for the manager, who
will have insufficient capacity for supervision of employees or for coordination of the work that is
performed. It also reduces the opportunities for promotion, which in turn may reduce motivation
In addition to the effect of the overall number of layers, one may also consider how tasks
are allocated to roles at different levels of the organization. Jaques (1990; 1989) argued that the
challenge is not one of centralization versus decentralization, but of ensuring that tasks are
allocated to the appropriate level of the organization. His key assumption was that people at
different levels hold roles with varying degrees of authority, and as individuals, also have
different degrees of “working capacity” (i.e., competence). The roles at different levels also
require a different time horizons: A CEO may have a time horizon of 10 to 20 years, while a
front line worker may only have a time horizon of hours or days in terms of the work tasks that
are being performed. Based on this rationale, he concluded that people at a low level of the
organization will not have sufficient political weight, long term view, or coordinating skills to
On the other hand, employees at the bottom of the hierarchy are usually closer to the
field (e.g., customer, suppliers) and thus have “local knowledge” that may be valuable in solving
a problem or performing a task. Authors who use agency theory as a starting point have often
23
framed this as a delegation problem (i.e., whether a principal (manager) should delegate a
decision to an agent (employee) at a lower level in the organization (Robinson & Stocken, 2013).
The conclusion is that there is a basic trade-off between better utilization of local knowledge
versus the risk of potential loss of control, assuming that the agent may not share the principal’s
interests. This issue is in turn associated with the steepness of the organization. The reason why
an employee at low level of the organization might be better able to make a decision is not only
related to the fact that he or she has local knowledge, but also that it is difficult (and takes time)
to transmit this information from the bottom to the top of the organization. Colombo &
Delmastro (2004) found that managers in organizations with many layers are less likely to
delegate decisions, because they have better access to the relevant information themselves (a
organizational complexity” in the literature. But it seems clear that managers believe that a high
number of layers represents organizational complexity. Some scholars (e.g., Colombo &
Delmastro, 2004) have also operationalized (vertical) organizational complexity as having a high
number of levels. It is clearly a rather rudimentary definition, because it does not take into
account the size of the organization, the tasks and work processes, and communication systems.
excessive number of layers, considering such contextual factors. If not, the concept is more akin
to complicatedness rather than complexity as defined by Suh (2001) (In the two preceding
“vertical” complexity is not too different from “horizontal” complexity (related to the functions
24
analogous to functional conflict or coupling, and the difficulty of transmitting information
vertically is analogous to coordination costs arising in work processes where sub-units at the
Design principles. As we have seen, the key goal should not be too minimize the
number of layers but to identify the appropriate number of layers, and avoid overlaps with
regards to roles and responsibilities between subordinates and superiors. The principles for a
“requisite organization” advocated by Jaques (1989) are intended to fulfil this purpose. He
proposed that roles should be arranged vertically according to task difficulty. Moreover, he
proposed that that each employee should be supervised by a manager working at one “stratum”
higher than themselves (in terms of working capacity and time horizon), to allow the manager to
provide appropriate context and direction for his/her employees. In practice, this means that one
needs to identify the tasks and the roles at each level in the current organization, and consider
whether each level adds value, and whether there is sufficient capacity for supervision of
One also needs to take into consideration the competence of the employees, the
particular tasks that are performed in the unit in question and how they are related to the tasks
performed elsewhere. A supervisor will have more capacity (and can handle a larger span of
control) if the employees are highly competent, perform similar tasks, and have few
interdependencies toward each other and toward people in other sub-units. He or she will have
less capacity (and will need to have a lower span of control) if employees are less skilled,
perform dissimilar tasks, and have many interdependencies that require the manager’s
25
DISCUSSION
Organizational complexity usually has multiple underlying causes. For this reason, one
highlighted three perspectives, all with their roots in systems theory, but nonetheless based on
different assumptions about the causes and consequences of organizational complexity (Table 4
and 5).
Specific design principles can be derived from each perspective that may be the basis for
interventions in organizations (Table 4). The principles are independent of the unit of analysis:
One may use the principles to simplify a department, a business unit, or an entire firm. As with
other types of organizational interventions, the first step would be to document the current
organization, assess the current degree of complexity, and identify the underlying sources. Data
would typically come from interviews with managers and employees, but electronic surveys may
also be used to identify interdependencies (e.g., Worren, Soldal & Christiansen, 2017). The
subsequent step would be to create new designs that lead to simplification in accordance with the
criteria listed in Table 4. This is partly a creative process where one generates alternative
solutions (in this case, organizational models) that may satisfy the requirements (i.e., the firm-
specific functions (design goals) and more generic design principles such as those listed in Table
4 and 5). However, analytical and algorithmic methods are gradually being introduced even in
this field. With regards to functions, one may use systematic methods to identify functions and
assess their relationship (see Worren, 2018b, for a suggested methodology). With regards to
work process interdependencies, algorithms have been proposed that can be used to cluster the
interdependencies and identify a more optimal grouping of roles or sub-units (e.g., Yu, Yassine,
26
Table 4. Key assumptions and design principles related to each perspective
27
Table 5. The main characteristics of complex versus simple organizational designs according to the three approaches discussed in the article.
Efforts at fulfilling one functional requirement negatively affects A role or sub-unit can work to fulfil its corresponding FR without
the ability to fulfil another functional requirement (i.e., negatively affecting the ability to fulfil other FRs.
functional conflict or coupling)
Information
processing
The formal boundaries are not aligned with the work process The formal boundaries are aligned with the work process
interdependencies – there are frequent or critical interdependencies – the most frequent or most critical
interdependencies across units. interdependencies are contained within units.
Stratification
The formal reporting relationships are not aligned with the The formal reporting relationships are aligned with the required
required roles at different strata. Some employees report to roles at different strata. Employees report to managers who are
managers who are at the same stratum as themselves in terms of one stratum higher than themselves in terms of working capacity
working capacity or authority. and authority.
When combined, the three perspectives constitute a broad and holistic view of
organizational complexity. However, the three perspectives have only limited value when used
separately. The main contribution of the functional perspective is to highlight the linkage
between unit mandates (functions) and specific organizational arrangements and to sensitize
effectiveness. But the functional perspective is mainly applicable to the horizontal relationships
(between roles or sub-units at the same level) and does not contain design principles for
of a hierarchy. The information processing perspective focuses on lateral work processes and
coordination costs related to information exchange and mutual adjustment. It is relevant for
understanding how interdependent roles may be integrated into sub-units, but ignores the
potential of functional conflicts (which might suggest a need for separation rather than
integration of roles and activities). Finally, the stratification perspective is useful when
considering the vertical structure and determining the number of management layers. But it does
not explicitly link the organizational model to the unit mandates (functions) or incorporate
organizational structure may have a highly complex governance model (e.g., Worren &
Scholderer, 2018) One can imagine that a reduction in the number of management layers might
increase rather than decrease complexity (defined according to the functional and informational
processing perspectives). This may happen, for example, if managers are forced to combine
(potentially incompatible) roles and take on “two hats” at the same time, such as a line manager
organization with self-contained units, that is, units that are focused on a key function or task,
that are relatively independent, and that possess the required authority and resources to produce a
whole product (or component, deliverable). This idea is far from new; it was introduced in the
scholarly literature 60 years ago, by March & Simon (1958). However, increasing complexity
has made it even more relevant today. It is, essentially, a prescription for enhancing the degree of
organizational modularity1 (Sanchez, 1995). The three perspectives may be used in conjunction
to operationalize this term: We may define a modular organization as one that minimizes
within and minimizes the number of interdependencies between sub-units, while nesting roles
and sub-units into a hierarchical system where larger units (business units, divisions etc.)
constitute more inclusive clusterings of units. Scholars have identified a number of benefits from
having an organization with modular or self contained units. It should lead to higher
productivity, given that it implies a reduction in coordination costs (Kilmann, 1983). It should
lead to higher effort, because key performance indicators and incentives can be tied to
measurable outcomes (Zenger & Hesterly, 1997). It should also lead to increasing innovation, as
the units are allowed to introduce local improvements to their work processes and /or product
1
I here focus on one aspect of modularity, the creation of semi-independent units, and ignore others (such as
substitutability and recombination of modules).
30
Linkages between the three perspectives
Although the three perspectives represent different facets of complexity, there are also
interrelationships between them. If there is functional conflict (or coupling) between roles or
sub-units, it leads to a need for coordination (marked with an a) in Figure 4). For example, a
situation where two sub-units have overlapping or conflicting mandates will trigger a need for
frequent information exchange in order ensure coordination between the sub-units and negotiate
how to best divide their efforts (The reverse is not the case, however: Work process
interdependencies between sub-units do not necessarily create a functional conflict. Units may be
interdependent from an operational point of view, while being independent in terms of unit
mandates or functions. The fact that one unit is dependent on another unit for certain inputs does
not in itself create a functional conflict (Worren, 2018b).) There are also some relationships
between the stratification and the information processing perspectives. The more (and stronger)
interdependencies there are between sub-units at one level, the more managerial capacity that is
needed at the next level in order to coordinate between the sub-units (the relationship marked
with a b) in Figure 4). Moreover, an excessive number of management layers tends to create
overlaps and conflicts between roles at different levels, which is analogous to functional conflict
(coupling) (the relationship marked with a c) in Figure 4). There is also a relationship the other
way: Functional conflict (or coupling) between sub-units at a lower level usually requires that
decisions related to the sub-units’ activities are placed at a higher level of the organization than
what would otherwise be the case (the relationship marked with a d) in Figure 3). For example,
other subsidiaries (e.g., because they have overlapping sales territories), the responsibility for
31
making pricing decisions needs to centralized rather than delegated (e.g., the decisions need to be
made by the headquarters rather than the subsidiaries themselves). Finally, there is a relationship
organization, means that information from the bottom of the organization must be transferred up
and down across more layers, leading to potential distortion of information and overload of
managers (the relationship marked with an e) in Figure 4). In sum, the relationship between span
of control and organizational complexity is not straightforward: Additional layers may on the
one hand contribute to increased complexity. But on the other hand, managerial capacity is also
needed to sort out interdependencies between and within sub-units. This suggests that delayering
needs to occur simultaneously with regrouping of sub-units. It is difficult to remove layers and
create a flat organization if one does not first create more modular or self-contained sub-units.
32
CONCLUSION
we need to develop frameworks that can help leaders in managing complexity and in simplifying
their organizations. In this article, I have shown that specific design principles can be derived
from three approaches – the functional, information processing, and stratification perspective.
The three perspectives are complementary, and should be combined when assessing the existing
one enhances modularity – one reduces conflicts and interdependencies and creates more
independent and self-contained units. In the realm of engineering, modularity is seen as the only
viable response to reduce the complexity of physical products and software programs. It might
REFERENCES
Ackoff, R. (1971). Towards a system of systems concepts. Management Science, 17(11), 661–
671.
Atherton.
Aken, J. E. van. (2004). Management Research Based on the Paradigm of the Design Sciences:
The Quest for Field-Tested and Grounded Technological Rules. Journal of Management
33
Argyris, C., & Schön, D. A. (1978). Organizational learning (14. print.). Reading Mass.:
Ashkenas, R. N. (2010). Simply effective : how to cut through complexity in your organization
Banet, A. G., Piehl, D. W., & McCann, J. (1976). Organizational Distance: A Concept for the
Analysis and Design of Organizations. Group & Organization Studies, 1(4), 496–497.
Browning, T. R. (2001). Applying the design structure matrix to system decomposition and
Clark, K. B., & Baldwin, C. Y. (2000). Design Rules: The power of modularity. Cambridge,
Collinson, S., & Jay, M. (2012). From complexity to simplicity Unleash your organisation’s
high-performance-organisation
Couto, V., Caglar, D., Plansky, J., & Recorded Books, I. (2016). Fit for growth a guide to
Cross, R., & Gray, P. (2013). Where Has the Time Gone? Addressing Collaboration Overload in
Daft, R. L., & Lengel, R. H. (1984). Information richness: a new approach to managerial
34
information processing and organizational design. Research in Organizational Behavior, 6,
191–234.
Day, G. S. (2006). Aligning the organization with the market. MIT Sloan Management Review,
48(1), 41.
Economist Intelligence Unit, E. (2015). Taming Organisational Complexity — Start At the Top.
127–33.
Eppinger, S. D., & Browning, T. R. (2012). Design structure matrix methods and applications.
Ethiraj, S. K., & Levinthal, D. (2009). Hoping for A to Z While Rewarding Only A: Complex
28–36. [Link]
207.
Greenwood, R., Diaz, A. M., & Li, S. X. (2010). The Multiplicity of Institutional Logics and the
Gresov, C., & Drazin, R. (1997). Equifinality: Functional equivalence in organization design.
Huckman, R. S., & Zinner, D. E. (2008). Does focus improve operational performance? Lessons
from the management of clinical trials. Strategic Management Journal, 29(2), 173–193.
35
IBM. (2010). Capitalizing on Complexity. Summers, NY.
Academic/Plenum.
Jacobs, M., Vickery, S. K., & Droge, C. (2007). The effects of product modularity on
27(10), 1046–1068.
Jaques, E. (1989). Requisite Organization: The CEO’s guide to creative structure and
Kilmann, R. H. (1983). The costs of organization structure: Dispelling the myths of independent
8(4), 341–357.
Kuhn, D. (2011). Delayering and Firm Performance: Evidence from Swiss firm-level Data.
March, J. G., & Simon, H. A. (Herbert A. (1958). Organizations. New York: Wiley.
Meier, K. J., & Bohte, J. (2003). Span of control and public organizations: Implementing Luther
Michael Novinson. (2017). DXC Technology trims four layers of management in bid to save
US$1 billion - Services - CRN Australia. Retrieved December 22, 2017, from
[Link]
save-us1-billion-470483
36
Morieux, Y., & Tollman, P. (2014). Six simple rules : how to manage complexity without getting
complicated.
Morris, J., & Farrell, C. (2007). The “post-bureaucratic” public sector organization. New
organizational forms and HRM in ten UK public sector organizations. The International
Nuñez, A. N., Giachetti, R. E., & Boria, G. (2009). Quantifying coordination work as a function
Pache, A.-C., & Santos, F. (2013). Inside the Hybrid Organization: Selective Coupling as a
1001.
Piran, F. A. S., Lacerda, D. P., Camargo, L. F. R., Viero, C. F., Teixeira, R., & Dresch, A.
(2017). Product modularity and its effects on the production process: an analysis in a bus
2331–2343.
PwC. (2006). 9th Annual Global CEO survey: Globalisation and Complexity; Inevitable Forces
in a Changing Economy.
Rajan, R., & Wulf, J. (2003). The Flattening Firm: Evidence from Panel Data on the Changing
Rechtin, E. (1991). Systems architecting : creating and building complex systems. Englewood
Rizzo House, R. J. & and Lirtzman, S. I., J. R. (1970). Role Conflict and Ambiguity in Complex
37
Robinson, L. A., & Stocken, P. C. (2013). Location of Decision Rights Within Multinational
Ashkenas; R. & Bodell, L. (201.). Seven Strategies for Simplifying Your Organization.
16(S1), 135–159.
Sanchez, R., & Mahoney, J. T. (1996). Modularity, flexibility, and knowledge management in
product and organization design.(Knowledge and the Firm). Strategic Management Journal,
Skrastins, J., & Vig, V. (2015). How organizational hierarchy affects information production.
Stea, D., Foss, K., & Foss, N. J. (2015). A Neglected Role for Organizational Design: Supporting
Suh, N. P. (1990). The principles of design. Oxford series on advanced manufacturing. New
Suh, N. P. (2001). Axiomatic design : advances and applications. The MIT-Pappalardo series in
Torsten Lichtenau, Smith, J., & Horrock, S. (2015). Tackling Complexity: How to Create Simple
38
and Effective Organizations - Bain & Company. Retrieved December 22, 2017, from
[Link]
[Link]
Ven, A. H. Van De, Delbecq, A. L., & Koenig, R. (1976). Determinants of Coordination Modes
Wallin, L., Pousette, A., & Dellve, L. (2014). Span of control and the significance for public
sector managers’ job demands: A multilevel study. Economic and Industrial Democracy,
35(3), 455–481.
Welch, J., & Welch, S. (2006). Winning : the answers : confronting 74 of the toughest questions
Williamson, O. E. (1983). Markets and hierarchies : analysis and antitrust implications : a study
in the economics of internal organization. New York; London: Free Press ; Collier
Macmillan.
Wong, C. A., Elliott-Miller, P., Laschinger, H., Cuddihy, M., Meyer, R. M., Keatings, M., …
Szudy, N. (2015). Examining the relationships between span of control and manager job
Worren, N. (2017). The matrix as a transitory form: the evolution of FMC technologies 2001–
Worren, N. (2018b). Organization Design: Simplifying complex systems (2nd ed.). Abingdon,
39
UK: Routledge.
Worren, N., Moore, K., & Cardona, P. (2002). Modularity, strategic flexibility, and firm
23(12), 1123–1140.
Worren, N., & Scholderer, J. (2018). The hidden matrix: Reporting relationships outside the
formal line organization. In Strategic Management Society conference, Oslo, June 7-9,
2018.
Worren, N., Soldal, K., & Christiansen, T. (2017). Organizing from the ground up: Developing a
DSM based tool for organization design. In 17th International Dependency and structure
Wulf, J. (2012). The Flattened Firm. California Management Review, 55(1), 5–23.
Yu, T.-L., Goldberg, D. E., Yassine, A., & Chen, Y.-P. (2003). Genetic Algorithm Design
Yu, T.-L., Yassine, A. A., & Goldberg, D. E. (2007). An information theoretic method for
209–222.
40