A decision theory perspective on complexity in performance measurement and management

PMM and decision theory in an age of complexity

One of the seminal works of decision theory is Simon (1947), “Administrative Behaviour: A study of decision-making processes in administrative organization”. This broke with the prior scientific management approach of Taylor (1911) by showing that managers could not perfectly control organisations because they lacked perfect knowledge. In recent years, decision theorists (French, 2012; French et al., 2009) have found the Cynefin framework (Kurtz and Snowden, 2003; Snowden, 2000; Snowden and Boone, 2007) to be a useful summary of the role of simplicity and complexity underlying organisational decision making. It is a sense-making framework that provides a typology of knowledge to assist managers in interpreting the nature of the context of various decision problems. The nature of the framework is shown in Figure 1 and Table I. Four domains are described where the level of knowledge about an external context is outlined, along with recommendations as to the appropriate way to respond.

Whilst applied in various fields from information technology to sociology (French, 2012; Herington et al., 2015; Pelrine, 2011), no consideration of the Cynefin framework has been found in the PMM or wider operations management literature. However, a direct similarity is found with the performance alignment matrix of Melnyk et al. (2014), a guiding framework for managers to consider strategy alongside PMM when operating in turbulent and volatile external environments. To resolve the problems faced by managers in reformulating strategy and PMM, the performance alignment matrix addresses how organisations relate to their external environment and how well the internal PMM aligns with the strategic goals of the organisation, plus how well they respond to the external business environment. Further definition and exposition is provided in the background literature section below.

The performance alignment matrix is a framework regarding the use of measurement that acknowledges the nature of complexity in organisations “that have multiple levels of decision-making and diverse operating contexts” (Melnyk et al., 2014, p. 182). Furthermore, Melnyk et al. (2014) describe one of the axes of the performance alignment matrix in terms of certainty. Certainty means being able to give specific outcomes, but under conditions of turbulence or fluctuations in the external environment, uncertainty prompts a flexible approach, stating objectives in general terms, not specific ones.

Decision theory, and in particular the succinct summary of sense-making in diverse contexts provided by the Cynefin framework, addresses the influence of organisational hierarchy and operating context on decision making (Figure 2). The types of sense-making provided in the Cynefin framework are also distinguished on the basis of the certainty of the decision maker; what is certain or known (and thus formalised into standard or instinctive operational responses), through what is knowable (after investigation via classical scientific method or expert analysis), through to what is only retrospectively knowable, due to complexity, or unknowable, due to chaos, reflects a similar spectrum.

Furthermore, both the performance alignment matrix and Cynefin are concerned with the practical issues of decision making in real-world applications. The similarities between the two are thus readily apparent, and correspond closely (see Table I). Given that the academic origins of each are different, there is a potential for deepening the theoretical foundations of PMM via an attempted synthesis of the two, providing a novel contribution to the PMM field through elaborating the existing theory.

The research question is thus stated as:

The contributions of this paper are threefold. First, we bring the decision theory concepts to PMM, and from our literature searches it seems we are amongst the first to do so. Second, we scrutinise such concepts through an empirical study entailing multiple case studies, and find supporting evidence for the inclusion of decision theory in PMM. Finally, we illustrate the cases using the Cynefin framework, and show how this might assist practitioners in understanding and responding to unpredictable environments when they measure and manage performance in their organisations. The role of organisational culture, defined in terms of dominant logic (DL), is central to shaping this response.

The structure of the paper is first to outline the various concepts being considered and provide clear definitions. A conceptual framework diagram is provided (Figure 3) that links the various elements. The research methodology and accompanying research design to explore the applications of PMM alignment using decision theory is then described. Findings and data analysis are then described, followed by a discussion of the findings and implications for future research and practical application.

Challenges in existing PMM systems

Harkness and Bourne (2015) discuss complexity as a barrier to PMM in terms of two aspects, the first of which is the complexity of the external environment. This is described as dynamic not static and as an “open system” not a “closed system”. Second, measuring performance is seen as being hampered by the factors including ambiguity, lack of control, unpredictability and unintended consequences. Acknowledging the presence of these as a challenge to PMM prompts deeper research into the origins of these concepts and hence an interdisciplinary bridge to equivalent discussions in the adjacent areas of study. This is referred to by Denyer et al. (2008) as “reciprocal synthesis”.

By incorporating decision theory into PMM, a deeper understanding is possible. A relatively recent summary of the decision theory field and its historical development is provided by French et al. (2009). The barriers to effective PMM noted by Melnyk et al. (2014) echo the definition of bounded rationality established by one of the seminal figures in decision theory – Herbert Simon. Bounded rationality is an essential part of the theory of business administration, complementing and contrasting with neo-classical economic theory of rational decision making (e.g. Taylor, 1911), because in practice, decision makers in organisations find, “their decision making processes are molded by limits on their knowledge and computational capabilities” (Simon, 1947, p. 20).

For PMM, bounded rationality provides a constraint on performance measurement in terms of the accuracy, availability or relevance of data. Performance management may then be affected by bias, where the decision is affected by cognitive factors on the perception of data and what it means. One further effect, studied by agency theory (Ross, 1973), is that of gaming the metrics in order to manipulate rewards offered by management (Moynihan, 2008). In other words, manipulating the performance through metrics adjustments in a way that triggers rewards or prevents the triggering of punitive responses by management. The actions of employees can thus become motivated by working to the PMM and the structures it creates, rather than to the traditional goals or objectives of the organisation. This may result in creation of wrong behaviour and mind-sets in an organisation where employees over-perform or under-perform to meet the targets against the assigned metrics set by management. Tett (2009) provides an account of this phenomenon in major banks during the lead up to the 2008 financial crisis. Such issues are common topics in the study of behavioural decision making (French et al., 2009).

The problem of too much focus on metrics in the approach to PMM (Bititci et al., 2012) are analogous to those discussed in the literature on systems theory (Forrester, 1958; Kumar and Kumar, 2016), which is not referenced in the PMM literature, but that have become explicit in the decision theory literature (discussed below) and the operations and supply chain management literature. Notably, it is the divide between general systems theory, dating from Forrester in the 1940s, and that of complex adaptive systems (Choi et al., 2001; Nair et al., 2009; Pathak et al., 2007), which is illustrated in the Cynefin framework. Rather than saying that everything is complex, Cynefin says that sometimes things are complex and sometimes they are not, and it is managerial sense-making of this context that is important.

One of the important question raised by PMM researchers (Bititci et al., 2012; Kennerley and Neely, 2003; Kennerley et al., 2003) was how to develop and maintain a dynamic PMM system over time in the climate of market volatility and turbulence. Kennerley et al. (2003) proposed a PMM maturity model that encapsulates the role of external factors such as market volatility and regulations, and internal factors such as culture, people and process in effectively managing and developing company’s PMM system that truly represents the organisation’s changing requirement. The paper addresses the “what” questions linked to dynamic PMM but had limited discussion on “how” to adapt and operationalise the new measures in a highly turbulent environment. However, this work could be considered as one of the early papers to debate the need for dynamic PMM.

Kolehmainen (2010) further picks up this theme to draw insights from a large telecommunication company and comments on the role of individual managers in adapting the PMM in a dynamic market environment. Flexible and strategically aligned PMM will allow different parts of the system to work together effectively and empower individual managers to adapt to turbulent market conditions. Such a system moves beyond command and control measures to measures that promote organisational learning (Kolehmainen, 2010; Micheli and Manzoni, 2010).

Pekkola et al. (2016) address the need for a dynamic PMM system called for in the literature (Bititci et al., 2012; Garengo et al., 2005; Kennerley and Neely, 2003; Melnyk et al., 2014; Nudurupati et al., 2011) by proposing a framework for flexible performance measurement system design for small to medium enterprises operating in a turbulent environment. Here, the core permanent measures of the framework are cost/profit centric and support measures aligned with the strategic goals. The supportive measures need to be flexible and dynamic to change as the organisation changes its strategic goals and vision when operating in the turbulent market conditions. However, this framework is too simplistic and lacks integration with decision theory, although it is a good starting point for small to medium enterprises to develop a flexible and robust PMM system.

As discussed by Melnyk et al. (2004, p. 212), the failure to respond to a dynamic environment causes stresses between the organisation, the external environment and the organisational strategy, which can become a source of conflicts between a firm and its customers or suppliers. The direct fit between environment and PMM has received less attention in the literature. Melnyk et al. (2014) address this by defining the “performance alignment matrix”. This is a newly created framework that categorises specific and general outcomes and specific and general solutions, which relate to the degree of certainty that defines management setting of outcomes (i.e. specific vs general outcomes) and specific approaches the organisation adopts to deliver the outcome (i.e. solutions).

Melnyk et al. (2014) classifies the performance alignment matrix as shown in Table I, above: measurement-driven management (specific outcome and solution), outcome-driven solutions (specific outcome and general solution), assessment-driven management (general outcome and solution) and solution-driven outcomes (general outcome and specific solution). The performance alignment matrix highlights that a failure to achieve alignment, or “fit”, between the internal PMM system and the external environment is a major barrier to effective PMM. Whilst providing a strong response to issues of predictability and unpredictability in PMM, such ideas are also present in decision theory fields and in particular the Cynefin framework, which is discussed in the next sub-section.

Decision theory and Cynefin framework

As described below, decision theory is one relevant area of business and management studies to examine complexity, uncertainty and unpredictability, alongside strategic management (Mintzberg and Waters, 1985; Mintzberg and Westley, 2001). In their account of the different branches and the evolution of decision theory, French et al. (2009) highlight the Cynefin framework (Kurtz and Snowden, 2003; Snowden, 2000, 2002; Snowden and Boone, 2007) as a useful and effective summary model (see Figure 1). Cynefin is a Welsh word (pronounced ku-ne-vin) roughly meaning habitat, including social and cultural elements as well as environment. It “signifies the multiple factors in our environment and our experience that influence us in ways we can never understand” (Snowden and Boone, 2007, p. 71).

The framework thus concerns the subtle perception of a decision context and is a typology of knowledge in relation to decisions, based on the nature of their context (French, 2012; French et al., 2009). It acknowledges the difference between the structured and predictable decision contexts, where managers can exert traditional command-and-control, and the unstructured and unpredictable decision contexts where they must either allow order to emerge from a group-oriented level, or attempt to impose order on apparent chaos. Figure 2, from French et al. (2009), links the epistemology of decision context shown in the Cynefin framework with the decision authority in a conventional organisational hierarchy (shown on the left side). PMM can sit at the level of tactical and operational decisions, ideally aligned with the higher level of strategic decision making.

The two sides of Cynefin (Figure 1) reflect a divide in epistemology (the philosophy of knowledge); the divide between rationality and bounded rationality defined by Simon (1947) in relation to management decision making, reflects this, and deliberately contrasts with Taylor (1911). As shown in Figures 1 and 2, the Cynefin framework consists of four domains, grouped into two halves of structured and unstructured domains. When a decision context is sufficiently structured and stable, it enables quantitative analysis. When it is characterised by dynamic change, plural and contested definitions, or non-linear complexity (Lorenz, 1963; Prigogine and Stengers, 1984), the space is unstructured (Kurtz and Snowden, 2003; Snowden, 2000). This offers a means to determine where it is possible to make rational decisions or where bounded rationality prevents this, and thus alternative methods are needed. The structured is further divided into simple and complicated domains, and the unstructured is divided into complex and chaotic domains.

Each domain suggests responses, detailed in Snowden and Boone (2007), that managers should take when faced with the different decision spaces defined by each domain. This responds to the concept of fit, concerning how well-aligned internal processes are to the external environment. The lines between each domain are curved to emphasise that the borders between the domains are contextual, and that the figure is not a graph with two axes marking quantitative variables. The Cynefin framework is effectively a typology of knowledge management, called a social ecology of knowledge (Snowden, 2000), in order to consider different approaches to organisational knowledge.

The four domains are also intended to illustrate that misalignment can occur when the external context of a decision may shift from one domain to another in a potentially dynamic fashion. The shift from a simple-structured domain to an unstructured-chaotic domain is for instance illustrated by a line symbolising a breaking wave on Figure 1. A simple bureaucratic system may cease to function effectively if the underlying assumptions shift and the context becomes increasingly structured or that the structure starts to break down into turbulence. It may become gradually more complicated, as additional requirements are added to a project for instance. Increasing the numbers of stakeholders involved in a project can result in a breakdown of shared perception, and this divide can result in an otherwise stable and structured context becoming less structured. These correspond to the concept of fit and subsequent description of the performance alignment matrix in Melnyk et al. (2014), which states that how a PMM responds to change is affected by organisational culture. PMM is thus both “a technological process and a social one […]” (Melnyk et al., 2014, p. 3).

Managerial sense-making and response is central to Cynefin and echoes the strategic management concept of dominant logic (DL) (Bettis and Prahalad, 1995; Prahalad and Bettis, 1986). This is described as the dominant mindset or view of the world possessed by top managers that shapes their decision making, and through their shaping of organisational culture, affects the decision making of others in the firm. DL is a schema based on the beliefs established as a result of employees’ experience within given firms and industries. Prahalad and Bettis were concerned with the role that DL plays when a firm in one industry acquires a firm in another industry and misunderstandings result, leading to underperformance. The perceptual, cultural and sense-making aspects of DL relate to how decisions are made in practice. This is important to understanding the three aspects of strategy, PMM and organisational culture mentioned by Melnyk et al. (2014) in terms of fit. In this paper, the Cynefin domains are taken as descriptions of DL, providing a decision theory perspective on organisational culture, which can then be applied to the issue of PMM and fit. Figure 3 expresses a conceptual framework covering the relationship between these key concepts.

Awareness of an organisation’s performance, captured via data, enables control, and as such the ability of managers to direct organisations towards meeting strategic requirements. Yet, if the context is unstructured as a result of unpredictability, then this approach is problematic. Cynefin suggests that under unstructured contexts, an approach based on responsiveness instead of forecasting, or emergence instead of control is required. PMM should reflect this dynamic, but the issue appears under-theorised in the literature, as noted by Melnyk et al. (2014) or Harkness and Bourne (2015), which address the role of complexity as a problem for PMM, but less so on how to respond to it based on the existing theory in this space.

Mapping the performance alignment matrix with the Cynefin framework

The performance alignment matrix framework is found to have a useful precursor in Snowden’s Cynefin framework. Introducing this framework and its associated conceptual underpinning helps provide a greater theoretical depth to the topic of PMM. This answers recent calls for deeper understanding of the role of volatility, uncertainty, complexity and ambiguity (Bennett and Lemoine, 2014) and how they are addressed as aspects of the external business environment, interpreted (via sense-making) in different ways by the internal organisational culture.

As stated by Melnyk et al. (2014), “for PMM to be effective it has to fit the environment in which it operates” (p. 183). Similarly, the Cynefin framework is built on an empirical observation; “Wise executives tailor their approach to the complexity of the circumstances they face” (Snowden and Boone, 2007). A range of sources of unpredictability, including volatility, uncertainty, complexity and ambiguity, have long been recognised as characteristic of manager’s perception of the external environment (Bennett and Lemoine, 2014; Simon, 1947). However, the Cynefin framework offers valuable insights in showing that such unpredictability is not always present. Many circumstances are readily predictable, and hence the decision makers “approach” should be tailored to the circumstances of the external environment in order to be effective – there is no “one-size fits-all” approach.

The failure of sense-making (Browning and Boudès, 2005; Weick, 1995) in relation to changing circumstances described by Cynefin is again similar to what Melnyk et al. (2014) call a “lack of fit” or misalignment. Notably, in defining the performance alignment matrix as a framework to describe alignment between PMM and the external environment, the authors cite precursors in strategic management; Venkatraman and Camillus (1984) describe strategy in relation to organisation theory’s open systems view where, “strategy [can] be conceptualised as a pattern or stream of decisions taken to achieve the most favourable match or alignment between the external environment and the organisation’s structure and process” (Venkatraman and Camillus, 1984, p. 516). This alignment, or fit, is thus a phenomenon that has been conceptualised over time in the management literature, particularly in strategy (Chorn, 1991; Henderson and Venkatraman, 1993; Walter et al., 2013). Both the Cynefin framework and the performance alignment matrix seek to conceptualise the phenomenon but do so from slightly different backgrounds: strategic management and organisational performance in the case of PMM, and knowledge management and decision making in the case of Cynefin.

The PMM concept of fit, as illustrated in the performance alignment matrix in Melnyk et al. (2014), can be elaborated using the more detailed explanations and managerial responses described in the Cynefin framework. Such decision theory concepts are empirically examined through case research into the implementation of corporate social and environmental responsibility practices via sustainable operations and supply chain management (SOSCM) and related PMM. Seven firms in different sectors are investigated to find how DL and the external environment are aligned and the subsequent alignment of PMM around SOSCM. The findings lend support to the adoption of the Cynefin framework as a useful means to understand how organisations should perceive and respond to the external environment. The next section provides more information about the method adopted for this research.

Case 1: electronics firm

Case 1 is a leading designer of electronic components for popular consumer devices. The large majority of the firm’s employees were engineers. The external environment was relatively stable and simple, with incremental improvements in product performance delivered year on year with little radical innovation. The DL was clearly structured and rules-based, shaped by scientific processes, so corresponding to Cynefin Domain 2.

SOSCM issues had become highlighted in the sector, with recent regulations imposed via the US Dodd-Frank Act. SOSCM issues had thus been introduced as a customer requirement. However, given the nature of the firm and its position in the supply chain, this was outside of the firm’s expertise or ability to influence (Q1.2). It was not an element included in the technical specifications that they passed up the chain to their suppliers in manufacturing and assembly. Hence, the response to SOSCM at the time of the interviews was bureaucratic and procedural (Domain 1), focussed on the legal liability of disclosure, requiring suppliers to confirm compliance, but having no mechanism of auditing or enforcement. The regulation resulted in minor changes in specification that reduced the potential liability but did not engage in supplier capacity building, or extending their engineering expertise to consider SOSCM issues more fully (Q1.1-Q1.5). Under the performance alignment matrix (Table I), this approach represented “measurement-driven management (specific outcome and solution)”.

Case 2: FMCG manufacturer

By contrast to Case 1, Case 2 had an SOSCM policy fundamental to the founding of the firm some 20 + years ago. The firm is an international manufacturer of ethical and ecological consumer goods. The interviews found a very strong organisational culture based around principles designed to maximise SOSCM performance in order to align with pro-ethical/pro-environmental customer values (hence, the strategic position of the firm). Notably, while Domain 2 processes existed in the lab and factory floor, the firm claimed insufficient resource for Domain 2 analytics in other aspects of management, including marketing and forecasting (Q2.3). Instead, the DL was clearly defined by Domain 3 logic (Q2.1, Q2.2). Characteristics of this were a decentralised approach to decision making inherent in PMM processes. A wide range of employees could be called in to work on decisions outside of their job roles via a principles-based approach to decision making (Q2.1, Q2.2, Q2.3). Firm strategy was essentially to align its performance with the ethical and ecological values of their consumers, establishing a form of quality strategy where exemplary SOSCM performance was fundamental to competitive advantage. Under the performance alignment matrix, this approach compared to the “assessment-driven management (generic outcome and solution)” domain where employees are given freedom to take decisions without too many constraints. However, while the DL was well-aligned to PMM, the firm stated that it lacked the resources to conduct Domain 2 analytics as some of its larger competitors did. Despite a niche position, because the sector of fast-moving consumer goods is mature and essentially stable, the Domain 3 position may be aligned to strategy, but misaligned to the external environment given the superior performance possible from a Domain 2 capability aligned with a stable competitive environment.

Case 3: restaurant chain

Case 3 is a large international restaurant chain with a strong, values-focussed culture intended to create the atmosphere of a family-owned restaurant despite a large size (Q3.1-Q3.2). The DL is Domain 3 as decentralised decision making and values of personal responsibility are central. With a strategic position of low cost but requirement for fresh produce (reducing the potential for inventory to smooth out variability in customer numbers) purchasing decisions were decentralised to the branch level. Local, granular knowledge and the instincts of branch managers were assumed to be required; knowing that a local sports team was playing on a particular day would prompt expectations of a busy day, rather than centralised ordering on the basis of forecasting via extensive Domain 2 data analytics.

Like Case 2, this firm lacked the Domain 2 analytic capability of a dedicated operational research department, and although it also had a principles-based, decentralised culture for decision making, it had this for very different reasons to Case 2, based more on the nature of the external environment as being characterised by high variability, than on resource constraint and strategic need for customer values-alignment (beyond the atmosphere of restaurants).

Two issues of SOSCM were explored; the first was a plan to reduce vulnerability to supply chain disruption. This was prompted by the EU horsemeat contamination scandal of 2015 but then applied to vulnerability to future climate change impacts. The inherent unpredictability of future disruptions did not affect investment in such initiatives because the Domain 3 DL of the firm had a high tolerance of ambiguity as to the business case. Such initiatives were supported by the board on the basis that if it sounded like the right thing to do, then it should go ahead (Q3.2).

The second SOSCM issue was the introduction of government carbon taxes (via the Energy Savings Opportunity Scheme). This was addressed as a calculation from the accounts department where total spending on electricity and fuel was converted into the associated carbon footprint and related government levy. This is a low-level, Domain 2 analytic process, or advanced, Domain 1 bureaucratic process. The policy did not however prompt a reduction in organisational carbon pollution as perhaps intended as to do so would require the firm to have a detailed data architecture, recording the operational consumption of every different location of the business. The size of the levy did not warrant the cost of developing a PMM system to target cost-effective reductions in energy consumption. This would be a more sophisticated Domain 2 undertaking, requiring expert analysis of a large number of variables. Instead, the carbon tax was seen as a cost of doing business (Q3.3). Latter firms in this study maintained a similar position towards this scheme. Misalignment can thus be considered in the SOSCM PMM as this is imposed by the government, as a broadly Domain 1 bureaucratic undertaking with some expert analytic Domain 2 capability needed, whereas achieving improvement in carbon performance would require an extensive and costly Domain 2 analytic approach involving the creation and study of potentially large sets of data to determine optimum performance output balanced against particular types of energy efficiency intervention (Q3.3).

While both Cases 2 and 3 had conventional, cost-based performance measures, it is notable that both were tolerant of ambiguity in decision making. This relates to the definition of a Domain 3 DL regarding level of certainty, providing some empirical validation of the Cynefin framework.

Case 4: high street finance firm

Analytic sensitivity to profitability was seen in Case 4, a stock market-listed international finance firm. The firm exhibits a Domain 2 DL as decisions were typically made on the basis of a complicated series of internal policies and regulatory standards, requiring bureaucratic responses and levels of expert analysis (Q4.1). PMM around SOSCM were well implemented and had been for many years, showing alignment between this aspect of PMM and the DL. Strong alignment between PMM for SOSCM and some strategic objectives could be seen in, for instance, energy efficiency programmes helping to improve productivity and profitability by reducing waste of energy. Similarly, local sourcing helped with local economic development, which improved reputation.

One other major SOSCM initiative was impacted by unpredictable volatility. A clear belief by the firm and its suppliers of linear energy price inflation (Q4.2) was affected by a sudden, unexpected drop in the price of oil a few weeks after one interview. A purely economic rationale for carbon reduction expressed in the interview was then undermined by the impact of this energy price fall on the anticipated payback periods for investment in energy efficiency measures. This represents a Domain 2 analysis becoming misaligned to an unpredictably changing external environment. Such issues of price volatility are perhaps inevitable factors of business, so alternative sense-making may have been cautious of such inevitability. In Case 3, the DL tolerated ambiguity about future paybacks, reinforced by the principles-based decision making. The key issue is not only how swiftly the firm responds, including how the PMM responds, but how the firm might anticipate such fluctuations and prepare accordingly.

A second key finding from Case 4 relates to a misalignment between customer perception of the business, the firm’s strategic goals around profitability, and the implications for DL. The firm was selected on the basis of its large size, and sophistication of its PMM systems, as a polar contrast to the small size of the previous three firms (as required by principles of theoretical sampling for high variation). However, a serendipitous discovery raised unprompted during the interviews was that the firm sought increased responsiveness to customer needs by introducing an alternative DL (Q4.1). This aimed at shifting from a Domain 2 structured, rules-based approach to a decentralised, principles-based Domain 3 logic.

Targets imposed by the former PMM were considered responsible for rising levels of customer dissatisfaction, which undermined profit targets and associated long-term strategic goals. Loosening the PMM system by shifting to a Domain 3 logic meant allowing customer-facing staff more discretion over decisions on the basis of their intuition about possible reputational impacts, rather than the pressure to hit certain department level performance targets. At the time of the data collection, this change management programme in the DL was at an early stage. Future research could examine the success of this, given the potential challenges of changing an organisation’s DL.

Case 5: construction contractor

Case 5 is a major contractor in the construction sector who showed a very strong Domain 3 DL, particularly in relation to project management and health and safety. Because construction sites are constantly changing on a daily basis and schedules and workflow are impacted by unpredictable issues like the weather, in ways that factory or office work is not, they require that project managers are able to respond to frequent change (Q5.1, Q5.3). They are also very dangerous places to work; from 2001-2011, 760 workers were killed on UK construction sites, compared with 448 British Army soldiers killed fighting overseas over the same period (HSE, 2016) (Q5.2). As such, a very strongly decentralised decision making culture had developed where people follow common sense principles instead of (blindly) following prescriptive rules.

The strength of this Domain 3 DL resulted in an unprompted account of the dangers of the balanced scorecard PMM system by major clients. Here, the misalignment between PMM and strategy resulted in serious problems for clients’ real estate portfolios; purchasing of land (deemed a performance metric) continued when macro-scale demand declined very rapidly after the 2008 financial crash. The misalignment resulting from this volatility has had a lasting impact on the real estate sector and the performance of the firm and its competitors in the sector (Q5.5).

In terms of SOSCM, the misalignment with PMM systems was also seen with the adoption of a government-promoted eco-standard for building design. The Building Research Establishment Environmental Assessment Method (BREEAM) system had become a widespread extension of updated mandatory building regulations affecting the environmental performance of buildings. But whilst building regulations were a simple, procedural and scientific measure (requiring expert analysis and so a Domain 2 activity), the BREEAM system consisted of a wide range of potential environmental features assigned scores via a more simple, bureaucratic Domain 1 system. Architects and contractors had to work with this Domain 1 system, even though the act of designing and constructing a building included many Domain 2, and Domain 3 activities. This antagonism was addressed in various ways, such as an architect firm becoming a certified BREEAM assessor so as to best balance the needs of the bureaucracy with the less mechanical and more aesthetic, qualitative or context-specific aspects of building design (Q5.4).

Theoretical implications of this are that while environmental performance of UK construction has been subject to substantial legislation for more than ten years, misalignment between prescriptive demands and the best-quality outcomes for buildings are clear. Future research could explore the nature of the misalignment in this sector in relation to BREEAM from the perspective of the Cynefin model of PMM described here.

Case 6: heavy manufacturer

Case 6 is a large, international manufacturer of construction products. SOSCM initiatives have become strongly aligned with the organisation’s strategy and PMM systems. As an engineering-based firm, a Domain 2 logic was dominant (Q6.1) and two large projects in SOSCM were investigated. First was the attempt to adopt the widespread use of life cycle analysis (LCA) on products. This involved a process of auditing environmental performance to create environmental product declarations (EPDs), a formal certification scheme. Possessing an EPD was rewarded by a point on customer supplier selection processes (a bureaucratic, Domain 1 activity). Yet, problems were seen in creating Domain 2 analysis of a product’s whole life cycle environmental impacts. With no regulatory or industry-level standard, the processes of LCA are highly variable. Furthermore, manufacturing a similar product out of different materials means highly non-comparable methodologies, assumptions and criteria. LCA is thus found to be a plural and contested Domain 3 undertaking (Q6.2).

Misalignment is thus found in the use of LCA as a PMM. Instead of becoming a Domain 2 system where an accurate and unambiguous analysis between two different suppliers is possible to determine which has the better environmental impact, LCA is in practice in a Domain 3 state. Meanwhile, in practice, because supplier selection processes award a point on the basis of having an EPD but not on any particular performance specified in the EPD, then LCA has become a Domain 1 undertaking. Establishing LCA as a Domain 2 system would require extensive regulatory standards and formal methodologies, independent assessment and auditing processes.

A second example of SOSCM in the PMM of the firm was on initiatives to reduce the carbon footprint of the firm. Two factors in the external environment impacted on this. First, international competition meant a strong downward cost pressure (Q6.3), meaning that energy efficiency measures were encouraged provided that they had a net cost reduction over a short payback time. Therefore, sustainability (implemented via SOSCM practices) as part of a strategic position of quality was problematic. Without such aspects being assisted by mandatory performance measures possible given the globalisation of the market and role of existing technical performance standards (that did not mandate environmental or social performance), the company’s competitive position had to adopt a strategy of low cost.

It was clear from customer tendering processes and weightings given to product specifications against cost, that price was the main decider, not sustainability criteria (or even issues such as product longevity or additional benefits such as service agreements) (Q6.3). This raises many questions for internal and external alignment. The practice of the firm suggested that PMM had become aligned to the new strategy to advance sustainability performance, but this strategy was becoming misaligned with the external competitive environment, and so the PMM for SOSCM refocused around the now better aligned strategy of low cost.

An additional instance of misalignment around SOSCM was found in discussions on the potential for changes to energy supply to low carbon alternatives. The highly unpredictable nature of the UK Government’s policy on this at the time was seen as undermining any potential for justifying investment in certain types of new plant (Q6.4). This contrasts with the example in Case 4, where there was an assumption of a predictable growth in energy price justifying investment over a given payback period. In Case 6, the larger scale of the energy consumption (given a manufacturing rather than services firm) meant the need for more substantial investment, more sensitive to government energy policies. However, anticipated unpredictability in this policy (rather than anticipated predictability in commodity price) prevented capital investment. Taken as a Domain 4 context, the responses to this unpredictable context in Cynefin are to either wait to see if order reappears over time or to try to impose order (Domain 4 left alone to Domain 3, or imposed to Domain 1). An example of trying to impose order may have been to lobby government to set clear new rules in place (Domain 4 to Domain 1 or 2). In practice, the firm waited without making a decision, delaying progress on this aspect of environmental performance.

Case 7: chemical processing

Finally, Case 7 is an engineering firm involved in the large-scale chemical processing of materials. The firm ostensibly operates as a services firm rather than manufacturer as its clients retain ownership of the materials undergoing processing. The firm described its activity as essentially very simple, albeit still in an expert area of chemical engineering and so is determined to have a Domain 2 DL (Q7.1). It only provides a small range of services but does so at a large volume, providing to a substantial share of global markets. SOSCM practices and associated PMM are regulated by government in environmental terms but extended into areas of corporate responsibility via investor requirements such as Global Reporting Initiative standards. As such, SOSCM is well established into key performance indicators (KPI) and embedded into PMM (Q7.2, Q7.3). In contrast to Case 1, this was also a strong engineering firm, and operated according to a clear Domain 2 logic, but it was one that had fully implemented SOSCM into its PMM processes. Two instances of alignment were also found. First, that an accident affecting a major customer impacted global inventory levels, as the production slowed, available supply of materials increased and so demand for their core product was reduced. This unpredictable event meant a change in the external environment, impacting firm strategy by reducing the market price. However, the PMM in relation to SOSCM was not affected, although there were strategic implications for the firm.

Second, when building new facilities, the firm’s stakeholder engagement capability needed to align with the local community concerns. This is a Domain 3 undertaking, and the approach taken of full transparency and open days at existing facilities is designed to improve the general public’s understanding of the business and its operations. This can be theorised as an example of enabling community sense-making so as to move the decision context of the community from Domain 3 plural interpretations (such as whether or not the factories had a risk of pollution or accident) to a common understanding of the high standards for safety and environmental management. This case was deemed the final one as a point of saturation seemed to be reached.

SOSCM and PMM

SOSCM as PMM is a new undertaking for many organisations and a relevant area for future research in its own right. This study has focussed on SOSCM as an implementation of PMM primarily from the decision theory perspective via the Cynefin framework. One general proposition to emerge is that by virtue of its novelty to many organisations, SOSCM represents a challenge for both PMM and strategy.

Focussing on the PMM aspects, a failure of strategic alignment, or lack of fit, can be found in terms of internal alignment. This concerns the extent to which there is a disconnect between the strategic direction of a firm sought by senior management, and the ability of operational-level PMM processes to align. Strategy is a case of external alignment. Here there is a challenge faced by many organisations as to how ecological and ethical issues are aligned or misaligned with their core strategy. For some businesses, external environmental changes in customer expectations or legislative change can challenge the core strategy. Firms dependent on fossil fuel extraction for instance, or changing attitudes to commodities such as palm oil, may need a change to the strategy of the business, or resistance to the new demands being exerted, commonly via lobbying against regulation or the rhetorical communications work of greenwash.

An additional finding from the data was in relation to DL, SOSCM and ownership type. In Cases 2 and 3, different ownership types placed less pressure to have clear and unambiguous economic Domain 2 justification for any SOSCM activity. Case 4 by contrast used a clear quantitative economic rationale to justify carbon reduction measures, but the volatility of oil price undermined this economic argument shortly after the interviews were conducted. This may mean that economic decision making for issues such as reducing carbon emissions are misplaced, and instead moral decision making is more appropriate.

Both firms in this study practicing this form of decision making as part of their DL (Cases 2 and 3) were privately owned, not publicly listed on stock markets. This prompts a future research question on the influence of ownership type on DL. The hypothesis would be that firms under private ownership may find it easier to establish values-based DL than those listed on major stock markets. Given the latter’s need for quantitative, Domain 2 accountability to unknown investors challenges to this form of DL may represent a barrier to addressing the issues of sustainable and responsible business.

The nature of knowledge – the extent to which something is known or certain – is central to both Cynefin and the performance alignment matrix (as shown in Table I). Various aspects of decision theory (indeed, various decision theories), including the nature of advanced quantitative analytics – “big data” or “artificial intelligence” – have been subject to extensive consideration in academic literature dating back several decades. So too has the area of ethical decision making, and the role of judgement, intuition and moral principles. One core message of decision theory is that any form of decision support system, model, set of principles or rule of thumb, is merely a tool that the decision maker uses to help guide or inform them; the responsibility for the decision taken still remains with the decision maker/s.

Seen as a framework within decision theory, Cynefin places greater attention on the role of perception, and the significance of organisational culture as a shaper of that perception for decision-making via its DL. The performance alignment matrix, by contrast, addresses “outcomes and solutions” as problems to be solved rather than as decisions to be made. By examining the real-world case studies, a body of empirical data has been collected that seeks to understand how organisations relate to their external environment and how their PMM and strategy align. As Cynefin is a sense-making framework that addresses both predictable external contexts and unpredictable ones characterised by complexity and volatility (where cause and effect cannot be determined in advance), it also offers insight into the issue of complexity in PMM.

Looking at the use of PMM systems that concern the application of SOSCM into organisations, the experience of changes in PMM, changes in strategy and changes in external environment have been subject to context-rich, in-depth exploration. Theoretical explanation of this can then be stated in terms of the Cynefin framework. Underlying assumptions of linearity or predictability are the key. For instance, the performance of energy consumption as part of programmes to improve energy efficiency is an instance of PMM. This was commonly framed as a driver of cost saving on the basis of a common assertion (found in interviews in Cases 4-6) that energy prices would only rise.

A similar example of SOSCM is also seen in Case 6 where LCA fails as a process to drive improved ecological performance. Merely having a certificate scores a point as a Domain 1 activity in supplier selection decision processes. Despite trying to generate accurate data, in practice no Domain 2 structured-complicated decision analysis could be developed to establish a clear, rational calculation of the benefit of one product over another. Instead, the process is actually a Domain 3 context, where there is a high level of ambiguity and context specificity. To make LCA amenable to structured decision making, such as supplier selection decisions based on environmental performance, would require a greater structuring via standardization of methodologies, as well as other means of determining valid comparability, that may not be possible (Q6.2).

By contrast with the performance alignment matrix, the Cynefin framework makes the different contrasts more explicit and provides deeper explanation as to why in terms of knowledge management and sense-making. It thus provides an extension or alternative perspective on the same phenomenon and categories described in the performance alignment matrix.

The main contribution provided by the research presented in this paper is illustrating the role of awareness of an organization‧s context. Conditions of volatility, uncertainty, complexity and ambiguity can mean that strategy is forced to change quickly, and that PMM must then change equally fast at strategic, tactical and operational levels. The example of a client making write-downs because of the impact of the balanced scorecard as a PMM in Case 5 (again, ultimately caused by volatility in real estate, post 2008), demonstrates the huge and negative strategic impact that can result from a brittle PMM that fractures in relation to strategy and environment. Organizational awareness, attention, or sense-making are central to this.

While this paper has used the Cynefin framework to elaborate existing theory in PMM, notably the performance alignment matrix in the work of Melnyk et al. (2014), the wider academic field of decision theory, related to the themes in both Cynefin and the performance alignment matrix, provides deep foundations from which to further develop theory. Returning to development of the alignment concept in strategic management (Chorn, 1991; Henderson and Venkatraman, 1993; Venkatraman, 1989; Venkatraman and Camillus, 1984) provides further opportunities to elaborate the theory of fit in PMM, potentially revisiting the roots of the debate in contingency theory (Hickson et al., 1971).

Conceptual synthesis (Denyer et al., 2008; Tranfield et al., 2003) may extend beyond the exploration of Cynefin and the performance alignment matrix here. The concept of dynamic capabilities, for instance, maintains that changing core capabilities in a swift and responsive manner is vital for strategic success (Eisenhardt and Martin, 2000; Teece et al., 1997). In the operations and supply chain management context, the hybrid concept of “leagile” refers to the balance between lean and agile where the former suits stable, regular process flows in mass production, against the operational flexibility required for dynamic capabilities to stay responsive to changing customer needs and other external forces (Purvis et al., 2014). These parallels and precursors further validate the concept of balance between different contexts (suddenly changeable and unpredictable, requiring responsiveness, or stable and predictable, and hence mathematically optimisable such as via statistical process analysis).

This distinction echoes longstanding discussions in business and management scholarship. These include the development of decision theory by Herbert Simon (1947, 1957, 1959, 1960, 1962, 1972, 1977, 1978); also, the critique of classical management science and general systems theory by Checkland (1972, 1980, 1983, 2000), where structured contexts are referred to as hard systems, and unstructured are soft systems (Figure 2), characterised by an inability to determine the variables; similarly, the concepts of tame problems, messy problems and wicked problems (Rith and Dubberly, 2007; Rittel and Webber, 1973) address the distinction later defined in terms of predictable, ordered simplicity and unpredictable complexity in Cynefin. Similar precursors also exist in the philosophy of science, such as Weaver (1948) “Science and Complexity” and Popper (1965) “Of Clouds and Clocks”. These precursors are examples of reciprocal synthesis, where the same phenomena are conceptualised using different terminology and from different scholarly perspectives. This range of literature helps to validate the nature of the Snowden’s Cynefin framework (and indeed Melnyk’s performance alignment matrix and Chorn’s typologies for strategic alignment).