A system dynamics approach to the balanced scorecard: a review and dynamic strategy map for operations management

The balanced scorecard (BSC), a strategic performance measurement tool ideated by Kaplan and Norton (1992), has been widely adopted by organisations seeking to turn an organisation’s strategy into a set of comprehensive performance metrics across four key areas: financial, customer, internal processes and learning/growth. Its application extends beyond traditional corporate domains as research in operations management (OM) increasingly recognises its value in fostering a holistic and integrated approach to performance evaluation (e.g. Brewer, 2000; Hu et al., 2017, p. 669). In this regard, Kaplan and Norton (1996b, p. 21) also assume that the scorecard should incorporate the complex cause-and-effect relationships amongst performance drivers (lead indicators) and outcome measures (lag indicators) that describe the strategy’s trajectory.

However, developing and designing BSC and determining which performance drivers (lead indicators) and outcome measures (lag indicators) is an ongoing debate (Barnabè, 2011). Notable BSC critiques include its neglect of certain internal factors, ambiguity regarding the weightage of its perspectives and an oversimplified emphasis on linear relationships between perspectives (Awadallah and Allam, 2015). One fundamental criticism is the BSC’s reliance on unidirectional and static relations amongst its perspectives, leading to concerns over the assumed cause-and-effect relationships. The concept of causality in BSC is not extensively explained and, at times, unclear. Nørreklit (2000, pp. 73–74) even posited that the connections in BSC are more logical than causal. Such ambiguities in the BSC can pave the way for misaligned strategies and operational actions (Barnabè and Busco, 2012).

Departing from the above limits, this paper aims to answer the following research questions:

Accordingly, we undertook a systematic literature review of the BSC literature in OM journals. We obtained a final sample of 40 articles, which allowed us to identify performance drivers and outcome measures for each perspective. Then, the system dynamics approach and causal loop diagram were applied to detect and synthesise the relationships amongst performance drivers, outcome measures and substantiated perspectives. In this vein, it is worth mentioning that whilst there are several notable prior reviews of BSC literature (see Supplementary Table A1 for details), none consider the role of the BSC for OM as their primary focus nor apply system dynamics.

The primary contributions of this article include (1) determining the foremost performance drivers and outcome measures within each BSC perspective, (2) illustrating their interrelationships, (3) showing empirical evidence on dynamic causality between BSC perspectives through these performance drivers and measures and (4) integrating these loops to create a dynamic strategy map for OM. This map provides a comprehensive view of the multifaceted relationships amongst the four BSC perspectives in OM – representing an extended version of the map by Kaplan and Norton (1996a). In brief, according to the results of our investigation, the different performance drivers and outcome measures within and between the four BSC perspectives in OM relate systemically without a clearly defined hierarchy and not according to trade-offs in dyadic terms.

Section 2 explores the study’s theoretical foundation, covering OM, performance systems, BSC and system dynamics. Section 3 details our systematic literature review methodology. Section 4 presents outcomes with causal loop diagrams for each BSC perspective. Section 5 introduces our reimagined strategy map for OM, emphasising the dynamic interactions between the BSC’s perspectives. Finally, Section 6 outlines our academic contributions and managerial implications, acknowledging limitations and suggesting future research directions.

Operations follow a common process of turning tangible or intangible inputs (e.g. physical goods, information and experiences) into outputs; however, the specifics of these inputs and outputs may vary (Schmenner and Swink, 1998, pp. 100–101). Indeed, according to systems theory, “to achieve a target state of the output, the execution of any process requires resources to bring about changes in the state of the flowing elements […] Hence, in its most basic form, a process is modelled as a flowing element interacting with a resource. For example, if the flowing element is a tree trunk, changing into logs for the fireplace, the resources consist of a saw, an axe and human labour” (Dekkers, 2017, p. 118). A further component of the input-transformation-output process is the 'feedback information,' which controls the input-transformation-output process (Fowler, 1999, p. 184).

Within this context, OM orchestrates the interconnected elements of the input-transformation-output process (Barnes, 2018). It holds significance across the entire organisation, offering principles, concepts, approaches and techniques that grasp value for managers. In particular, this includes sourcing products and services from suppliers and ensuring their smooth delivery to customers, collaborating seamlessly with other organisational functions, providing a continuous operational flow and fostering internal efficiency (Radnor and Barnes, 2007).

By quantifying performance metrics, operations managers can set benchmarks, track progress and drive continuous improvement efforts, ultimately enhancing productivity, reducing costs and delivering better customer value (Bititci et al., 2012). Amongst several performance measurement systems developed to match internal operations capabilities with external market requirements, one of the most well-known and accepted performance measurement system tools is the BSC, established by Kaplan and Norton in 1992. The relationship between its four perspectives – financial, customers, internal and learning and growth – can be synthesised as follows. Learning and growth of employees’ skills and abilities allow the development of processes that may lead to increased effectiveness in internal operations. This, in turn, enhances the value provided to the customer, which is then converted into improved financial results (Kaplan and Norton, 1996a). Therefore, these perspectives allow companies to monitor short-term financial results whilst tracking the progress and performance of intangible assets that generate growth for future financial performance.

The four perspectives can be represented as an interlinked bottom-up hierarchy, called strategy map (see Figure 1), and their performance drivers and outcome measures should be linked in cause-and-effect relationships (Kaplan and Norton, 2004, p. xii). Operating as lead indicators, performance drivers are peculiar to a particular business unit, albeit oriented towards those universal objectives. Instead, the outcome measures, such as profitability and customer satisfaction, frequently operate as lag indicators. This is a practice-oriented distinction, but there is no insurmountable separation between lead and lag indicators. Instead, each business chooses which measures are lag and lead for itself according to its objectives, fields of competition, resources and so on. For instance, Kaplan and Norton (1996b, p. 21) consider employee satisfaction as a generally used lag indicator, whilst several field contributions depict employee satisfaction as a lead indicator of customer satisfaction (e.g. Macpherson, 2001, p. 17). Definitively, a genuinely working BSC should include and coordinate both performance drivers and outcome measures.

In this respect, studies have confirmed the key assumptions of the BSC and its validity as (1) deploying strategic intent into single objectives and measures linked to the four different perspectives (Otley, 1999, p. 375), (2) creating consensus about the strategy required to generate organisational integration (Lipe and Salterio, 2000, p. 285) and (3) integrating outcome measures (lag) and the performance drivers of outcomes (lead), linked together in cause-and-effect relationships horizontally within and between areas (Butler et al., 1997, p. 247). Accordingly, the BSC works not just as a measurement system, but integrates non-financial measurements in a strategic control framework to support the value creation process in organisations (Bourguignon et al., 2004, p. 115).

The characteristics of the BSC in terms of strategy execution and management also suggest potential benefits for companies in OM (Carmona and Grönlund, 2003). Indeed, the BSC can improve organisational activities by (1) translating strategy into operational goals, framing objectives comprehensively, (2) understanding relationships amongst performance drivers and outcome measures and (3) involving employees through a systematic evaluation of role clarification.

Nevertheless, the BSC has been criticised in terms of effectiveness and efficiency, in particular, its inability to move plans and strategies into action in a timely way (Barnabè, 2011; Bianchi and Montemaggiore, 2008), due to its difficulty in fully integrating the effect of dynamics within a system. In particular, Nørreklit (2000, p. 78) highlights the absence of cause-and-effect relationships between measures from the four perspectives. Although the connections are likely interdependent, Kaplan and Norton (1992) refer to finality, not causality. This suggests that the fundamental assumption underpinning the BSC needs to be refined.

To this end, a further step has been introducing a second- and third-generation BSC (Lawrie and Cobbold, 2004) based on the development of the strategy map initially proposed by Kaplan and Norton (1996a). Here, a diagram provides a deeper causal analysis amongst performance drivers and outcome measures, highlighting the value creation process by connecting strategic goals in relationships in the four BSC perspectives. Barnabè (2011, p. 468) emphasises that links between BSC perspectives are interdependent and illustrates the dynamic nature of the system based on the strategy map. We will advance this insight by proposing diagrams and a strategy map.

There have been several attempts to develop dynamic BSC across different settings within organisations (e.g. Bianchi and Montemaggiore, 2008). These have been primarily designed to consider the feedback loop approach and mainly tested against real-world data (Akkermans and Van Oorschot, 2005). These different practical developments, applying the tracking of dynamic relationships between the four BSC perspectives, may provide OM research with a relevant theoretical base to develop a dynamic strategy map for organisations. As suggested by Oladimeji et al. (2021), a structured system dynamics approach to performance management can help organisations strategically achieve their goals and support decision-makers by moving from a static to a dynamic view to show causality, systemic connection, time delay and interrelationships. In other words, BSC advancements based on a system dynamics approach aim to demonstrate that matching the traditional BSC architecture with system dynamics principles offers better support for strategic management decisions (Barnabè, 2011; Supino et al., 2019) and is also relevant to operational studies (e.g. Cunha Callado and Jack, 2015).

In 1961, the seminal book Industrial Dynamics by Jay W. Forrester brought to light the system dynamics approach ^[1]. The system dynamics approach is “a perspective and a set of conceptual tools that enable us to understand the structure and dynamics of complex systems” (Sterman, 2000, p. 6), it was advanced to help executives gain a more in-depth understanding of newly framed complex contexts. More recently, Dekkers (2017, p. 285) defined system dynamics as “an approach to understanding the behaviour of complex systems over time. It is mainly based on internal feedback loops and time delays that affect the entire system’s behavior. Generally, it is applied to analyse any dynamic system characterised by interdependence among elements, mutual interaction between actors and elements, feedback loops, and circular causality”. In other words, the system dynamics approach highlights a continuous view of the organisation that uncovers system behaviour and the structures underlying discrete decisions (Forrester, 1961). In particular, the system dynamics approach focusses on understanding the behaviours of complex systems, which are usually composed of components in circular interlocked relationships (Forrester, 1994). Such circular causality concerns the influence of component A on component B, which in turn influences a component C that affects the original component A, thus, determining a circular A-B-C relationship. To estimate the impact of a given corporate strategy, the system dynamics approach can help identify ex ante complex situations where unpredictable causes generate unplanned effects, allowing managers to handle those situations effectively (Barnabè, 2011). Methodological steps for applying system dynamics can be summarised as follows (Sterman, 2000; Shaik and Dhir, 2021).

1. Defining the problem/situation: defining the problem/situation and the system’s boundaries.

2. Identifying the variables and their relationship: defining the crucial component of the system and depicting the causal relations amongst them through causal maps that also identify positive or negative feedback and/or delays.

3. Modelling: using qualitative and quantitative tools, such as diagramming tools, stock and flow maps and causal loop diagrams to model system dynamics. Causal loop diagrams successfully depict system dynamic-based BSC (e.g. Bianchi, 2016; Supino et al., 2019); it consists of nodes (i.e. variables) and edges (i.e. causal links between variables). A positive link means two nodes change in the same direction; if the node in which the link starts decreases, the other node decreases and vice versa. A negative causal link means the two nodes change in opposite directions: if the node where the connection starts increases, the other node decreases and vice versa.

4. Verifying, validating and simulating the model: ensuring the validity of the system dynamics model for its predetermined use. For this purpose, the system dynamics model may be submitted to different types of validation tests, such as the classic statistic test, technical validation, or case studies (Barlas, 1989; Barnabè, 2011).

5. Analysing the results: understanding the future behaviour of the system and designing policies that can help the system to work better.

The system dynamics approach has also been advocated by Kaplan and Norton (1996a, p. 67) as suitable for BSC research: “the Balanced Scorecard can be captured in a system dynamic model that provides a comprehensive, quantified model of a business’s value creation process”. Identifying the dynamics amongst the four BSC perspectives can facilitate the systemic approach of OM and has implications for managers seeking to enhance organisational performance (Kaplan, 2009). In other words, the system dynamics approach for the BSC (1) enables understanding of interrelated dynamic relationships amongst performance drivers and outcome measures, (2) elicits mental models and sharing knowledge amongst organisational agents, (3) allows better dissemination of strategy to both managers and staff, (4) permits the identification of the potential consequences of management policies and (5) provides a better linkage between the performance measurement system and organisational strategy.

Nielsen and Nielsen (2015, p. 1), combining elements from traditional BSC with systems thinking, suggest a shift from a static to a system dynamics approach to the BSC. However, these authors do not consider the several dynamic relationships existing within and between the four BSC perspectives when implemented in the OM field – as also claimed in recent system dynamics and BSC contributions (Oladimeji et al., 2021; Tawse and Tabesh, 2022).

The systematic literature review method was chosen because, unlike traditional narrative reviews, it “locates existing studies, selects and evaluates contributions, analyses and synthesizes data, and reports the evidence in such a way that allows reasonably clear conclusions to be reached about what is and is not known” (Denyer and Tranfield, 2008, p. 671). In this article, such a rigorous and reproducible method is employed to shed light on the multiple dynamic relationships within and between the four BSC perspectives when implemented and used as a strategic tool in OM.

This systematic literature review follows authoritative guidelines (i.e. Tranfield et al., 2003; Denyer and Tranfield, 2008) and related applications (e.g. Cristofaro and Giannetti, 2021; Heinis et al., 2022). To transparently report the purpose, methodology and findings of our review, we adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement – a comprehensive guideline crafted by Page et al. (2021) that outlines the steps for selecting, assessing and synthesising studies in systematic reviews. The PRISMA guidelines include a five-phase flow diagram (see Figure 2). The five steps are (1) identification, (2) screening, (3) eligibility, (4) inclusion and (5) data extraction and analysis. The checklist includes items deemed essential for the transparent reporting of a systematic review; questions within the list are broadly divided into the following categories: title, abstract, introduction, methods, results, discussion and other information.

Consistent with Tranfield et al. (2003), we selected articles for the systematic literature review as follows.

1. Identification. The databases used for the literature search were (a) Business Source Premier (EBSCO), (b) ProQuest’s ABI/Inform, (c) Web of Science and (d) Scopus. Only peer-reviewed journal articles published in English were included. The research was not restricted to a given starting period; the end date was 1 September 2023. Selected articles were required to contain at least one of the following words within titles, keywords and/or abstracts: “balanced scorecard” OR “BSC” OR “scorecard” (similarly to other BSC systematic literature reviews, e.g. Hansen and Schaltegger, 2016). From this, 9,831 contributions emerged.

2. Screening. Duplicates from databases were eliminated at this stage, resulting in 3,511 contributions.

3. Eligibility. The following eligibility criteria were applied to focus on articles in the OM field and ensure the retrieved documents' relevance and quality. Only articles published in journals ranked in the field of “Operations and Technology Management” (OPS&TECH) or “Operations Research and Management Science” (OR&MANSCI), according to the Academic Journal Guide (AJG) 2021, were selected. Then, we eliminated articles not published in journals ranked 2, 3, 4 and 4* in the AJG 2021; Hristov et al., (2021). This step ensured a certain level of academic rigour for filtered publications. Moreover, to observe a relatively unbiased procedure, we also considered journals ranked 1 in the AJG 2021, whether contextually classified as A*, A, or B in the Journal Quality List (JQL) 2021; see Supplementary Table A3 for the comparison. After applying these eligibility filters, our sample consisted of 241 articles. These were fully read to ensure their alignment with the research objective. Following the approach adopted by Poggesi et al. (2016), the 241 articles were reviewed according to two quality assessment criteria: (a) theoretical robustness and (b) methodological robustness. For both requirements, all authors assigned scores to each article ranging from one (lowest value) to three (highest value). Articles with scores less than or equal to three were excluded from the sample. The inter-rater reliability for this assessment was high (Cronbach’s Alpha = 0.87). This phase led to a final sample of 39 articles.

4. Inclusion. The snowballing technique was adopted to consolidate the research outputs and one article was added. The final sample was composed of 40 articles.

5. Data extraction and analysis. For each article, we retrieved the following information: (a) authors, (b) journal, (c) year of publication, (d) type of article, (e) context of the study, (f) data collection procedure, (g) data analysis procedure, (h) main findings and (i-j) performance drivers or outcome measures [including the adopted key performance indicator (KPI)]. Note that indicators are selected two by two, thus considering their relationship and that neither a minimum nor maximum limit was applied to the number of coupled indicators retrievable from each article reviewed; (k) polarity of the relationship (“+:” or “-”) between indicator 1 and indicator 2; (l) main results associated with the indicators 1 and 2 relationship; (m) the direction of the relationship; (n) BSC perspective related to the relationship between indicators 1 and 2; (o) the notation within the related causal loop diagram; and (p) the strength of the relationship (“moderate” for relationships identified in one article amongst those reviewed, “substantial” for relationships identified in two articles, “strong” for relationships identified in three or more articles). See Supplementary Tables A2 and A4 

6. Data analysis. The system dynamics analysis was implemented according to the following four-step procedure (see also Supino et al., 2019):

7. Two authors read each article separately, reporting in a detailed worksheet (see Supplementary Table A4) the couples of indicators (indicators 1 and 2) for each article that substantiate OM influences, connecting each indicator to a specific BSC perspective according to its main area of impact;

8. The polarity, direction and strength of the relationship between the two indicators were identified, linking indicators in the feedback loop by causal connections that can be charged with a polarity to determine the type of effect, that is, a positive polarity (“+”) for a straight influence and a negative polarity (“−”) for an inverted relationship, helping to determine causal relationships between two indicators, for example, A positively influences B;

9. Considering the specific BSC perspective connected to the extracted and studied indicators, we aggregated, two by two, the indicators’ relationships, helping form a causal loop diagram for each BSC perspective by hypothesising each loop’s start and end from a specific outcome measure according to its potential causality (see Barnabè, 2011) and then classifying feedback loops into two types, positive and negative. Positive (or reinforcing, “R”) loops intensify the dynamics emerging within a system (i.e. if the cause increases, the effect increases above what it would otherwise have been, and if the cause decreases, the effect decreases below what it would otherwise have been). In contrast, negative (or balancing, “B”) loops balance the dynamics within a system (i.e. if the cause increases, the effect decreases below what it would otherwise have been, and if the cause decreases, the effect increases above what it would otherwise have been) (Sterman, 2000);

10. We aggregated causal loop diagrams to form a dynamic strategy map for OM by first identifying the articles that analysed relationships amongst indicators belonging to different BSC perspectives (which we term “outward relationships”), then analysing the relationships’ causality in terms of direction and polarity (see Supplementary Table A5), and finally, aggregating the already drawn causal loop diagrams considering those ‘bridging’ connections.

In this section, we depict the causal loop diagram for each BSC perspective through Figures 3–6. Graphically, outcome measures are written in uppercase, whilst performance drivers are in lowercase. Lowercase letters amongst indicators (i.e. notation; a, b, c, etc.) indicate the logical order of the relationships depicted. Moreover, the strength of the relationships (as reported in column p of Supplementary Table A4) is indicated through different thicknesses of the connections, namely a strong relationship between indicators is shown by the thickest line, a substantial relationship by a line of medium thickness and a moderate relationship by the least thick line.

The BSC learning and growth perspective includes objectives mainly expressed by employee-based indicators. These indicators were then retraced and connected, as extensively reported in Supplementary Table A4, then depicted within a dynamic system reported here through the causal loop diagram in Figure 3. The first reinforcing loop generated in this analysis (R1) is named “Organisational learning and growth loop”, which starts from and culminates in the outcome measure organisational culture (oriented to learning and growth), including three unidirectional and two reciprocal positive relationships.

Organisational culture incorporates essential elements, such as trust, commitment, collaboration and competencies, to create virtuous dynamics within the learning and growth perspective centred on people’s skills and tasks. Thus, central to an organisational culture (oriented to learning and growth) is employee training and empowerment and, consequently, employees’ technical and human improvements may be valued by the organisation’s members who, in turn, trust and reinforce that organisational culture (Albuhisi and Abdallah, 2018). Once a good level of employee training and empowerment is reached, it improves employee motivation (Akkermans and van Oorschot, 2005) as an enhanced positive psychological attitude toward their tasks strongly increases employee productivity. The positive influence of employee motivation on employee productivity is particularly valuable given its generalisability. Indeed, this relationship is detectable in different types of organisations, both large (Akkermans and van Oorschot, 2005) and small (Quezada et al., 2014) and in both service (Akkermans and van Oorschot, 2005) and product (Singh et al., 2018) organisations. Increasing employees' productivity means improving the company’s capacity to fulfil tasks requested and, moderately, even enhancing employee satisfaction and positive attitude toward the work done (Akkermans and van Oorschot, 2005). Also, organisational culture creates a stimulating work environment where employees generally feel more satisfied and optimistic. Employee satisfaction supports people expressing those positive attitudes and aims at developing the desired organisational culture to restart the loop (Decoene and Bruggeman, 2006).

The internal processes perspective identifies those processes where the organisation must excel in satisfying customers and shareholders over time (Kaplan and Norton, 1992); thus, procedures aim to deliver outstanding value to end users whilst maintaining the highest possible internal efficiency. Starting from and ending with the outcome measure internal and environmental efficiency, the causal loop diagram of the internal processes perspective – extensively explained in Supplementary Table A4 and depicted in Figure 4 – identifies three positive reciprocal relationships and five positive unidirectional relations within the second reinforcing circle of this analysis (R2), named efficiency loop.

The internal efficiency of organisations involves appropriately utilising both tangible and intangible assets, including information. Hence, improved interior and environmental efficiency may support timely and correct information flow, measured by information system efficiency (Okongwu et al., 2015). Vice versa, once the organisation is provided with an efficient information system, it can increase its final level of internal and environmental efficiency (Okongwu et al., 2015) as actors in the production chain encounter correct and timely information to reduce those misalignments usually associated with consumption and costs. The efficient information system may also benefit good cross-operational communication (Aliakbari Nouri et al., 2019), attempting to lead to cross-operational integration (Andersen et al., 2004). Indeed, all those activities that need to be completed with the participation of different operational levels (or departments) require prompt intersecting communications to pursue their integration (Andersen et al., 2004). Then, repeated interactions amongst operational levels (or departments) naturally create organisational routines over time (Andersen et al., 2004). Indeed, routines can generate standardised practices to be adjusted day-by-day to address delays and errors reduction, which is quantified explicitly as a reduction in the differences between the due date-actual date and expected delivery-actual delivery by Supino et al. (2019) in the case of an e-commerce implementation project. Reducing errors means fewer undelivered and returned product ratios, resulting in waste reduction (Brewer, 2000). In parallel, these efficient dynamics generate outstanding results in customer deliveries and waste reduction, determined as recycling rates by Reefke and Trocchi (2013), both determining positive levels of internal and environmental efficiency (Ferreira et al., 2016). In turn, delays and errors reduction is relevant to improving internal and environmental efficiency. In particular, reducing errors involves reducing waste, production cycle times and costs due to returned products, whilst reducing delays consists of reducing costs due to stretched delivery times (Chand et al., 2005). Once the firm has figured out these operational issues and reached a convincing internal and environmental efficiency level, the loop may restart with the reinforced mechanisms.

The causal loop diagram of the customer perspective comprises indicators creating the fundamentals that show customers the company’s value proposition (Kaplan, 2009). In balancing the customer satisfaction loop (B1) identified in Figure 5, three positive reciprocal relationships, five positive unidirectional relationships and one negative unidirectional relationship – each explained in detail in Supplementary Table A4 – start from and end with the outcome measure customer satisfaction.

Even though the content of the value proposed outwardly differs between companies, customer satisfaction most likely increases corporate reputation in customers’ minds. Vice versa, final users directly express the overall favourable assessment of the company’s reputation in the market through measures like customer satisfaction (Barnabè, 2011). Nevertheless, for the organisation, these sustained levels of appreciation are significant when stable economic relationships are achieved by retaining customers (Goharshenasan et al., 2022). Therefore, customer retention may be exploited by firms to reinforce corporate reputation further, still considering the positive influence that the latter has on the former (Nielsen and Nielsen, 2012). Companies want to obtain sustained sales from their relationship with their audience. Hence, a positive corporate reputation is built to pursue a self-reinforcing association with corporate brand awareness in the relevant market (Reefke and Trocchi, 2013). Indeed, the presence of a specific offer on the market likely allows the organisation to increase its market share, widely estimated by the percentage of a market segment served by the firm and then a more extensive awareness of the value proposition in the market (Hu et al., 2017). This allows the organisation to be more in touch with customers, the management of whom often relies on customer relationship management (CRM) systems. An effective CRM, namely keeping relationships with customers and analysing their preferences over time, positively affects customer satisfaction. In particular, customer satisfaction does not derive from the CRM effect on product-service quality, which is still debated (Okongwu et al., 2015), but rather from the CRM effect on the creation of reciprocally satisfying relationships with customers (Reefke and Trocchi, 2013). Nevertheless, CRM effectiveness may decrease when orders rise to high levels, especially if inexperienced and derived from different market segments (Okongwu et al., 2015). Customer satisfaction can ‘reopen’ the whole loop at this final stage.

This final perspective comprehensively evaluates the financial aspect dynamics. Hence, Figure 6 shows the fourth reinforcing loop in this analysis (R3), named the profitability loop with the highest number of connections (ten positive unidirectional, all explained in Supplementary Table A4). The loop’s beginning and end lie in the outcome measure profitability.

Improving the capacity of people and systems to operate efficiently is crucial to sustain outstanding economic results. Investments authorised within the R&D budget can assume a strategic relevance, as in the Swedish electrical engineering company considered by Nielsen and Nielsen (2012). Indeed, as R&D activity comprises creative and systematic work undertaken to increase knowledge stock and devise new applications of available knowledge, such companies generally systematically report lower variable costs (Aliakbari Nouri et al., 2019). Nevertheless, such innovation-driven changes do not always optimise the production process. Rather, it is conditioned to the firm’s characteristics and the industry where it operates. For example, the empirical outcomes of Lee and Kang (2007) offer a partial endorsement of the concept that the variety of innovation impacts productivity growth. Their findings point out that, in the short term, process innovation could yield more substantial strides in productivity compared to product innovation. This observation stems from the divergence in efficiency progression when deconstructing productivity growth into two key components: efficiency growth and technical growth. To elaborate, product innovation inherently involves the creation of novel products and revolutionary changes, which could impede efficiency growth to a greater extent than other forms of innovation due to the intricacies involved in product development and the necessary adaptations for “innovations.” Conversely, process innovation is directed towards minimising defects, shortening lead times, curtailing costs and addressing other factors, rendering it strongly oriented towards augmenting efficiency. Consequently, it significantly contributes to the enhancement of efficiency growth.

Expanding upon the insights outlined earlier, it becomes clear that the intricate interplay between various types of innovation and their ramifications for productivity growth holds profound implications. As underscored by Voelpel et al. (2006, p. 53), the character of innovation tied to R&D activities is undergoing a transformation from incremental to disruptive, from closed to open and is increasingly becoming network-driven. This shift towards companies embracing strategic connections, sharing knowledge and adopting transformative practices carries escalating significance. Therefore, organisations seek to embrace production process-cost efficiency (variable costs reduction) with their final intention to increase general profitability, and this relation is measured by residual contribution margin (Dror, 2008). The positive economic result increased the economic value added (EVA) calculation, which, assuming stability of the cost of capital, may contribute to enforcing corporate investment capacity (Hu et al., 2017). As firms are usually supposed to be part of a competitive economy, their investment capacity will be exploited to improve their fixed assets. More productive machines and bigger plants allow the firm to pursue revenue growth, generally determined as a positive difference between current and past revenues (Hu et al., 2017). In addition, such an increase in revenues could improve fixed cost coverage, which is compensation for the stable economic efforts of the firm (Supino et al., 2019). Then, as the contribution margin is the excess between the selling price of the product and total variable costs, the residual may be used to cover fixed costs (which are covered at the break-even point) (Cunha Callado and Jack, 2015). After the break-even point, an increase in the residual contribution margin can be translated into profit relevant to profitability indexes (Tjader et al., 2014), then restarting the profitability loop.

In this section, we propose an extension of the strategy map for OM. Consistent with Nørreklit (2000), we show that the relationship between the BSC perspectives is more likely to be interdependent. This map identifies dynamic relationships amongst the performance drivers and the outcome measures of the four BSC perspectives and reconsiders them as a system characterised by complex patterns (see Figure 7). The relationships between BSC perspectives are defined as “outward relationships”, thus generated by “outward indicators.” Lead indicators of one BSC perspective that influence other lead indicators of other BSC perspectives are represented in lowercase and fine arrows (see Figure 7). Amongst such outward indicators, we include an indicator originally considered an outcome measure (i.e. employee satisfaction and positive attitude) that, in Figure 7, takes on a performance driver function because it positively influences another performance driver (i.e. corporate reputation). The outward relationships amongst performance drivers are synthesised and represented through bold arrows that connect the four BSC perspectives' outcome measures. The arrows are coloured according to the causal loop diagram of the BSC perspective they start from. In total, 13 positive unidirectional relationships were found amongst performance drivers. In contrast, three reciprocal and one unidirectional relationship were found amongst outcome measures of different BSC perspectives (see Supplementary Table A5 for full description).

We note five outward relationships starting from the causal loop diagram of the learning and growth perspective. Employee satisfaction and positive attitude may be seen as an outward indicator of improving corporate reputation (from the customer perspective) by diffusion of positive employee experience to the business audience (Akkermans and van Oorschot, 2005). Then, the training offered to employees is connected to two critical outward relationships. Indeed, higher levels of employee training and empowerment, on both operational and relational abilities, enhance, respectively, quality deliveries (measured through delays and errors reduction from an internal process perspective) and relationships with customers (measured through CRM effectiveness from a customer perspective) (Chand et al., 2005; Okongwu et al., 2015). From the financial perspective, two more outward effects are produced by employee productivity. First, productive employees are better at using corporate fixed tangible and intangible assets, allowing the firm to raise fixed costs coverage related to their utilisation (Quezada et al., 2014). Second, productive employees speed up the processing of incoming orders, thus allowing new orders to be processed and sustaining revenue growth (Akkermans and Van Oorschot, 2005). The described relationships confirm the central role assumed by the learning and growth perspective within the structure and functioning of the BSC. The reinforcing outward effects are directed from the outcome measure organisational culture (from the learning and growth perspective) to all the other outcome measures, namely internal and environmental efficiency (from the internal processes perspective), customer satisfaction (from the customer perspective) and profitability (from the financial perspective).

Using waste reduction, the causal loop diagram of the internal processes’ perspective influences the causal loop diagrams of two other perspectives. First, as substandard products are typically wasted and reprocessed, reducing this phenomenon consequently allows for cutting time and reducing mistakes in deliveries – with final positive outcomes on employee productivity (learning and growth perspective). Second, reducing wasted and reprocessed products inevitably lowers costs (financial perspective), especially variable costs, because they are directly linked to the number of manufactured items (Brewer, 2000). Moreover, delays and error reduction move toward generating a positive reputation for the firm in the market (customer perspective) (Chand et al., 2005). The outward relationships from the internal processes causal loop diagram are synthesised by reinforcements from internal and environmental efficiency to profitability, organisational culture (oriented to learning and growth) and customer satisfaction.

From the customer perspective of the causal loop diagram, two outward influences are directed to two different perspectives. Since a positive corporate reputation is mainly built on reinforcing relationships between a firm and customers, such favourable consideration increases with the number of orders entered from sales. Therefore, higher sales and reputation improve corporate revenue growth (Reefke and Trocchi, 2013). From the customer perspective, the causal loop diagram generates outward relationships that the reinforcements may summarise from customer satisfaction (the customer perspective) to profitability.

The financial causal loop diagram gives rise to three outward effects, two from the customer perspective and one from the internal processes’ perspective. From a customer perspective, firms with a high capacity for investing obtain a double positive effect. Indeed, relevant investment capacity allows the firm to preserve independence from other business entities and be considered a leading player in the market, respectively increasing variables of corporate reputation and brand awareness (Hu et al., 2017). Regarding the internal processes causal loop diagram, performing R&D activities enables the organisation to find new ways to optimise the production process to reduce waste (Aliakbari Nouri et al., 2019). The outward relationships generated by the financial causal loop diagram may be condensed in the transitions from profitability to internal and environmental efficiency and customer satisfaction.

The systemic view of connections identified in the proposed loops and the extended strategy map offer relevant contributions to the OM literature from both a strategic and dynamic point of view. In other words, OM is not solely concentrated on transforming the input of raw material into goods and services – in a “siloed” way for which functions are vertical and disconnected from each other – but is at the centre of exchanges amongst other subsystems that should be considered when carrying out company activities efficiently and effectively (as is the aim of OM) (Adam, 1983, p. 366). This contrasts with the “sand cone” model (Ferdows and De Meyer, 1990), for which all four sustainable competitive advantages can be developed by following a particular sequence of strategic priorities. Whilst we recognise the empirical strength of the “sand cone” model, the proposed system dynamics relationships between the BSC perspectives discussed here adopt a less hierarchical view. At the same time, however, we cannot fully support the trade-off model of operational capabilities (Skinner, 1969), for which improving any one of the four basic manufacturing capabilities – quality, dependability, speed and cost – must necessarily be at the expense of one or more of the other three. The reciprocal relationships framed dynamically and systemically in this study underline how leveraging a variable in a BSC perspective (e.g. employee productivity) can differently impact other BSC perspectives (e.g. CRM effectiveness and revenue growth); that is, they do not always work as a trade-off.

Undertaking a systematic literature review of BSC articles published in OM journals, this study proposes dynamic relationships within and between the four BSC perspectives, when implemented and used in OM contexts, according to a system dynamics approach. We identify four loops, one for each BSC perspective, that describe the multiple dynamic relationships amongst the different performance drivers and outcome measures and their reciprocal influences (positive or negative) within each perspective. This, in turn, allows the creation of a dynamic strategy map for OM, which represents the relationships between the four BSC perspectives via interlinked loops rather than through an interlinked bottom-up hierarchy. This is the first study of its kind and theoretical and managerial implications can be drawn.

This study contributes to theory in several ways. First, our results provide the most representative performance drivers and outcome measures and their relationships within and between the four BSC perspectives in OM contexts. By doing so, our article identifies relationships amongst the indicators (perfomance drivers and outcome measures) and uncovers the dynamics and systemic interactions between them. It provides an advancement over more traditional, static views of performance measures and their interrelationships. In this vein, our literature-based findings (1) reinforce previous studies (e.g. Oladimeji et al., 2021), suggesting the need to move from a static to a dynamic view to help organisations better achieve their goals in the face of the renewed and ever-increasing contexts complexity and (2) gather empirical evidences to fill the gap – at least for the OM field – raised by Nørreklit et al. (2018) about the lack of empirical studies on causality between BSC perspectives.

Second, the provided causal loop diagrams and extended strategy map facilitate organisations in implementing a performance measurement system, making the connections between the different performance drivers and outcome measures explicit according to the system dynamics and addressing one of the main BSC criticisms, that is, its excessive focus on unidirectional linkages, simplistically exploited to linearly link measures across the four perspectives (Akkermans and van Oorschot, 2005; Barnabè, 2011). Indeed, our findings demonstrate the existence of dynamic and sometimes reciprocal linkages, thus implying the possibility that these links will change over time, along with internal (e.g. organisational dimension) and external aspects (e.g. sector trends) of the firm.

Third, we are neither fully adhering to the “sand cone” model (Ferdows and De Meyer, 1990) of operational capabilities nor the trade-off model (Skinner, 1969). Our approach is that the different outcome measures and performance drivers within and between the four BSC perspectives in OM relate systemically, without a clearly defined hierarchy and not always balancing trade-off effects amongst indicators in the organisation. BSC indicators’ connections work in loops; three out of four are positive, with just one that can be defined as balancing (i.e. customer satisfaction loop). This means the indicators’ trade-offs cannot be identified by looking at indicators in dyadic terms. This is consistent with the system dynamics logic and can be the subject of future empirical investigations.

Lastly, the findings of this article provide relevant theoretical support to the viable system model (VSM) (Beer, 1984). In particular, Beer (1984, p. 18) highlights that subsystem “three”, representing the controls and structures ensuring synergy amongst operations, is difficult for organisational members to be recognised. The results of this work confirm how the BSC is a useful tool to control the cohesion of operations, supporting organisational viability. Integrating the system dynamics approach to the BSC perspectives provides valuable support to the systematic control of operations, improving the effectiveness of subsystem “three” of the VSM.

Applying the proposed BSC dynamic strategy map for OM can pose challenges for practitioners, as its contribution primarily rests on a conceptual framework. The execution of models drawn from prior research is inherently complex, yet having a foundational framework can prove especially valuable in identifying the initial step for practical implementation. Whilst no universal “best way” exists, several optimal paths are conceivable, contingent on the organisation’s circumstances. These circumstances encompass internal facets such as the organisational life cycle stage, dimensions, culture, adopted strategies and external aspects like industry trends, historical context, technological advancements and societal shifts. However, the extended strategy map introduced in this study can serve operations managers by facilitating their comprehension of several key aspects: (1) the dynamic interrelationships within and between each perspective of the BSC; (2) the most relevant performance drivers and outcome measures within each perspective; (3) the reciprocal influences of different indicators within individual perspectives; and (4) the interconnections between the various perspectives themselves. Subsequently, the primary subsequent step for managers is to implement the provided loops and extended maps tailored to the unique operational context of their respective organisations.

Adapting the proposed strategy map to the specific context is imperative to ensure the performance management system’s efficacy. To achieve this, companies can follow a systematic procedure based on the performance measurement system literature (i.e. planning, action implementation and feedback and alignment) (Hristov et al., 2021) including system dynamics application steps reported in subsection 2.3 (i.e. identifying of the problem/situation, identifying the variables and their relationship, modelling using system dynamics software, constructing a stock and flow diagram, verifying, validating and simulating the model and analysing the results) (Shaik and Dhir, 2021). This systematic procedure may be essentially performed through the resulting three steps: (1) planning, which includes pre-evaluation of the problem, objective identification (i.e. variables and their relationships), strategy map development (i.e. modelling) and simulation building (i.e. validating the model); (2) action implementation (i.e. simulating the model); (3) feedback and alignment (i.e. analysing the results). By systematically following these three steps, organisations can tailor the strategy map to their specific realities, enhancing its alignment with their unique goals, challenges and industry dynamics.

Given the review nature of this work, the proposed strategy map for OM is not practically implemented, but only possible steps for its implementation are provided. Thus, scholars interested in OM and the BSC should enrich the proposed dynamic strategy map for OM by testing its practical validity and generalisability. In particular, future research could focus on the specific conditions under which the relationship between performance drivers and outcome measures is either positive or negative and determine if a trade-off model of operational capabilities can be verified under some circumstances. In addition, in line with the concept of “finality” (Nørreklit, 2000, p. 77), the multiple dynamic relationships within and between each BSC perspective is developed based on logic and not empirics. Further studies could use our results to empirically test these relationships, exploring how the presented strategy map affects the formation of operational capabilities in a co-evolutionary fashion (Cristofaro and Lovallo, 2022).

In this vein, and according to a system dynamics view, empirical analysis proving a statistical connection between the non-financial and financial indicators will be an important step in contributing to the BSC development. We suggest that BSC scholars build empirical models aimed at validating our results. For example, collecting data through a survey to quantify customer satisfaction on a Likert scale, thus creating a customer satisfaction index and statistically testing the correlation between this index and one or more profitability indexes [e.g. return on assets (ROA) and return on investment (ROI)].

Moreover, the BSC implies a nomothetic approach; this denotes that BSC and its derivatives, including the proposed strategy map, need to be better suited to specific instances caused by competitive heterogeneity and characteristics of industrial sectors. Accordingly, future studies may start from our generic strategy map to explore particular situations through intensive study of a single case to validate proposed relationships. This focus can refine existing theories and redefine their boundaries (see Mode 2 - Borrowing and Extending; Zahra and Newey, 2009).

These issues encourage idiographic and longitudinal research, such as case studies to test the indicators' relationships over prolonged periods. In addition, based on the collected data, statistical analysis can provide in-depth insights into the model’s validity. In particular, it potentially provides the evidence supporting specific relationships and the basis to overcome the assumption undermining Nørreklit’s finality (see Mode 3 - Transforming the Core, in Zahra and Newey, 2009).

In addition, the strategy map based on the BSC does not cover the intricacies of investments and resources allocated to research and development that may be dispersed and not just captured by budgets for R&D. Investments in R&D may lead to prestige or many other benefits (Jaruzelski et al., 2005). Future research could explore this phenomenon, extending, for example, the analysis of the R&D budget and R&D activity.

The other limits of this work relate to the systematic literature review protocol used. It may exclude some relevant literature (such as accounting) because it limits the data collection to articles published in selected OM journals. The results of our study can be replicated and extended beyond the OM field, selecting and analysing articles outside the operation discipline. Yet, as for all systematic literature reviews, our study is influenced by the heterogeneity of contexts (despite all of them dealing with OM), data collection methods and measurements.

In addition, the selection based on the ranking of journals should have been replaced with an assessment of the quality of evidence (see Dekkers et al., 2022, p. 140). For instance, an adaptation of the Newcastle–Ottawa Scale (NOS) could be considered. Also, the grey literature (books, book chapters, etc.) is not explored, which could be relevant given that publication bias is likely in business and management studies (studies into failures or deficiencies of concepts, theories and so on are limitedly published). Another limitation is the assumption that the strategic map is an adequate reflection (model) of actual processes. Also, reference models, such as the breakthrough model in Dekkers (2017), could provide a backdrop for extensions and validation.

These limitations have been consciously considered from the beginning and through the analysis phase, so it is reasonable to believe that the probability that excluded research contained information that would critically alter the conclusions reached has been reduced. Future research may take a quantitative approach to study the proposed relationships and consider external actors' influence on firm operations. This would help see BSC implementation in OM as even more “systemic.”