Data analytics use in financial due diligence: the influence of accounting and commercial logic

Abstract

Purpose

The purpose of this paper is to contribute to a better, empirically grounded and theoretically informed understanding of data analytics (DA) use and nonuse in accounting for decision-making. To that end, it explores the links between accounting logic, commercial logic and DA use in financial due diligence (FDD).

Design/methodology/approach

The paper reports the findings of a case study of DA use in the FDD practice of a Big Four accounting firm in Sweden (Pseudonym: DealCo). The primary data comprises semistructured interviews, observations and additional meetings. Institutional logics is mobilized as method theory.

Findings

First, accounting logic and commercial logic both drove and hindered DA use in DealCo’s FDD practice in different ways. Second, conflicting prescriptions for DA use existed mostly within commercial logic rather than between accounting logic and commercial logic. Third, accounting logic and commercial logic, as perceptual and conceptual filters, seemed to shape DealCo’s advisors’ understanding of DA and give rise to an efficiency-centric DA logic. This logic, in turn, as a high-level model of how to use DA in the context of FDD, governed DA use broadly.

Originality/value

The paper draws attention to direct and indirect links between accounting logic and commercial logic, on the one hand, and DA conceptions and use, on the other hand. It, thereby, advances prior theorization of DA use in accounting for decision-making.

Citation

Kastrup, T., Grant, M. and Nilsson, F. (2024), "Data analytics use in financial due diligence: the influence of accounting and commercial logic", Qualitative Research in Accounting & Management, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1108/QRAM-10-2023-0188

1. Introduction

Perhaps the primary implication of the notion of accounting for decision-making is that accounting practice of today must move ahead, with new thinking and new methods (Davidson and Trueblood, 1961, pp. 581-582).

More than 60 years ago, Davidson and Trueblood (1961) asserted that accounting needed to divert itself from its historical preoccupation with fiduciary and stewardship responsibilities, to support broader management decisions, and had “an obligation to take a significant part in the development of new quantitative information systems” (p. 582). Today, the accounting field is undergoing digital transformation (Arnaboldi et al., 2022; Möller et al., 2020; Quattrone, 2016) and is, again, caught in a debate around the need for new methods and new ways of thinking (see Bhimani, 2020; Bhimani and Willcocks, 2014). This is perhaps most evident in the area of data analytics (DA) and the calls for developing an “analytics mindset” in accounting (Mintchik et al., 2021; Richardson and Watson, 2021; Schmidt et al., 2020). DA, most commonly, refers to “extensive use of data, statistical and quantitative analysis, explanatory and predictive models […] to drive decisions and actions” (Davenport and Harris, 2007, p. 7). As such, it comprises many ideas and methods, e.g. exploratory analyses of unstructured data (Richins et al., 2017), that lie outside the scope of traditional accounting work, which has long revolved around visible and verifiable financial in- and outflows (Bhimani and Willcocks, 2014; Broadbent, 1998).

However, times are changing and DA is now seen as a must-seize opportunity for accounting professionals to create added value for their clients and organizations (including management) (Richins et al., 2017). For instance, Bose et al. (2023, p. 39) link DA to “offer[ing] additional value-added services,” Maisel and Cokins (2014, p. 63) associate it with “value and growth” and Nielsen (2018, p. 167) discusses its role in making “management accountant[s] a real value driver for the company.” Despite these prospects, by and large, DA adoption in accounting has been surprisingly slow (Casas-Arce et al., 2022). Indeed, albeit very limited, field research on DA adoption and use in accounting practice suggests that, so far, accounting professionals have realized only a small fraction of the supposed value of DA (see Buchheit et al., 2020; Schmidt et al., 2020). This seems to fit with the observation of “reluctant accountants with a traditional mindset” (Arnaboldi et al., 2017, p. 821) and a “rigid focus on traditional measures” (Agostino and Sidorova, 2017, p. 791). It is also in line with prior reports that (management) accountants’ DA activities remain focused on financial data analysis, and that external and operational data are not used extensively (Spraakman et al., 2021). In short, there seems to be a gap between the prospects of DA use, as described and discussed in large parts of the accounting literature, and the reality of DA use, as found in accounting practice.

As for why this gap exists, the above seems to imply the concurrent influence of competing prescriptions for what constitutes appropriate behavior concerning DA adoption and use – e.g. generating added value vs honoring accounting traditions – in accounting. More specifically, it seems to imply the concurrent influence of different institutional logics (Friedland and Alford, 1991; Thornton et al., 2012):

[R]esilient social prescriptions that enable actors to make sense of their situation by providing “assumptions and values, usually implicit, about how to interpret organizational reality, what constitutes appropriate behavior, and how to succeed” (Greenwood et al., 2010, p. 521, citing Thornton, 2004, p. 70).

This includes “commercial-professional logic,” which we call commercial logic, and “technical-professional logic” (Spence and Carter, 2014, p. 949), which we call accounting logic. The former specifies appropriate and meaningful behavior concerning the commercial aspect of accounting work; the latter specifies appropriate and meaningful behavior in relation to the technical issues embodied in accounting work (ibid.). These distinctions provide the conceptual basis for exploring how accounting and commercial logic influence DA use in accounting practice. This is done to contribute to a better, empirically grounded and theoretically informed understanding of DA use and nonuse in accounting for decision-making – the purpose of this paper.

Knowledge about DA use and nonuse and, in particular, accounting and commercial logic’s influence on it, is important for several reasons. On the one hand, it can illuminate the question of why DA is not used as extensively in accounting as in other business functions. On the other hand, it can help evaluate whether this lag is a matter of concern. Some have proclaimed that “the profession must embrace and adapt to a data analytics mindset or face extinction” (Schmidt et al., 2020, p. 167). Yet, perhaps, said lag exists for good reason and accounting would be ill-advised to embrace DA in similar ways to other professions. Against this background, invoking the institutional logics perspective and especially the distinction between accounting logic and commercial logic, promises new insights into accounting-DA tensions, DA’s business value in accounting and – closely linked – how accounting logic impedes or, perhaps, is implicated in DA-based value creation. Insights into these matters, in turn, can enrich current debates and also inform discussions about how to best approach DA adoption and use in accounting practice.

Empirically, the paper is based on a case study of the financial due diligence (FDD) practice of a Big Four accounting firm in Sweden (“DealCo”). FDD is an investigation into a company’s financial performance, positions and health in the context of mergers and acquisitions (M&A) that intends to support and facilitate M&A decision-making among buyers. Upon realizing that DA can significantly enhance FDD practice, the Big Four have invested in “deal analytics” and “deal technologies” and begun to digitally transform their FDD practices (Deloitte, 2021; EY, 2020; KPMG, 2021; PwC, 2021). Unlike auditing, FDD is not subject to regulation or oversight, which means that – for the most part – scopes reflect what clients desire and perceive as value-adds. At the same time, in selling FDDs, the Big Four prey on their long-standing reputation as elite players in the broader assurance field, especially in relation to financial information. Bad DA, e.g. based on erroneous data or flawed models, could tarnish this reputation. This adds an element of institutional complexity and makes FDD a highly suitable setting for investigating the concurrent influence of accounting and commercial logic on DA use.

The present study makes three interrelated contributions to the literature on digitalization in accounting, especially in relation to DA and accounting for decision-making (e.g. Bhimani and Willcocks, 2014; Casas-Arce et al., 2022; Knudsen, 2020; Möller et al., 2020; Schneider et al., 2015). First, the study highlights key links between accounting logic, commercial logic and the (non)use of DA tools in FDD practice. Examples include concerns around traceability and data integrity and considerations related to product quality and competitive capacity. Beyond that, it also draws attention to a second, indirect mechanism of influence. In DealCo’s FDD practice, accounting and commercial logic seemed to act as perceptual and conceptual filters that shaped DealCo’s advisors’ understanding of DA, i.e. what DA was and what it was good for, and, in effect, their use of DA tools. By showing that, our study heeds recent calls for theory- and field-based evidence on DA (non)adoption and (non)use in accounting for decision-making (see Casas-Arce et al., 2022; Schmidt et al., 2020). Second, by detailing the different ways in which accounting logic conflicts, and does not conflict, with DA use in FDD practice, it adds nuance, and an alternative viewpoint, to the debate around tensions between the traditional “accounting mindset” and the new DA approach (e.g. Agostino and Sidorova, 2017; Arnaboldi et al., 2017). Third, by showing the influence of commercial logic, in light of accounting logic, and drawing attention to conflicting prescriptions within commercial logic – in relation to DA use in FDD – the paper extends prior discussions about the business value of DA in accounting (e.g. Brands and Holtzblatt, 2015; Nielsen, 2018).

The remainder of the paper is structured in the following way. First, we offer a short literature review and develop central concepts (2). Thereafter, we describe materials and methods (3). Subsequently, we present the findings from the case study (4) and discuss them against prior literature (5). To conclude, we summarize the main contributions, note implications for accounting practice, highlight key limitations and suggest avenues for future research (6).

2. Theoretical background

In the following, we give a short review of the literature on DA in accounting (2.1) and develop the central concepts of this study, first and foremost, accounting logic and commercial logic (2.2).

2.1 Data analytics in accounting

DA draws on resources from three fields, that is, technology, quantitative methods and decision-making (Mortenson et al., 2015). Hence, it is neither a tool nor a technique but an approach to insight generation (for decision-making) in which data, tools and techniques/methods are used and combined in certain ways (see Delen and Zolbanin, 2018). First, in terms of data, DA often involves big data, i.e. high-volume, high-velocity and/or high-variety data (Vasarhelyi et al., 2015). Big data comprises structured and unstructured data and may be produced, respectively, collected inside (e.g. finance or operations) or outside of the organization (e.g. social media or national statistics). Common examples include geographic data, mobility data, sensor data, social media data and socio-demographic data, among others (Alles and Gray, 2016; Bhimani and Willcocks, 2014; Cockcroft and Russell, 2018). In a retail context, for instance, FDD providers could combine geo- and socio-demographic data to analyze store saturation and validate growth targets. Second, in terms of tools – hardware and especially software – DA requires four basic elements: infrastructure, data management, data analysis and information delivery; essentially, organizations need next-generation tools to store, process, analyze and communicate (insights from) large data volumes efficiently and effectively (see Rikhardsson and Yigitbasioglu, 2018). In concrete terms, this means accounting professionals must move beyond spreadsheet software and adopt newer, more powerful DA tools (Brands and Holtzblatt, 2015; Schmidt et al., 2020), either in the form of commercial packages (e.g. Alteryx, PowerBI), scripting languages (e.g. R, Python) or self-developed/proprietary tools. Third, in terms of techniques and methods, to fully harness big data, accounting professionals have to leverage new and more complex data analysis techniques (Appelbaum et al., 2017; Nielsen 2018, 2022). Contrary to the traditional accounting toolbox, which is geared toward description and diagnosis, much of the DA toolbox is geared toward prediction (Appelbaum et al., 2017). Thus, embracing DA may also require a shift in orientation from the past to the future (Nielsen, 2018). Doing the above, i.e. leveraging more data, new tools and more advanced techniques, is associated with creating more business value – by facilitating better decision-making – and making accounting more relevant (see Bose et al., 2023; Franke and Hiebl, 2023; Maisel and Cokins, 2014; Nielsen, 2018).

Despite these prospects, as noted in the introduction, (albeit limited) field research suggests that DA adoption and use in the broader accounting field has been comparably slow, especially in regard to more advanced applications, such as predictive analytics (e.g. Buchheit et al., 2020; Krieger et al., 2021; Spraakman et al., 2021). In particular, prior field research points toward hesitation concerning each of the central DA dimensions described in the preceding paragraph, namely, new (and big) data sources (see Agostino and Sidorova, 2017; Arnaboldi et al., 2017; Spraakman et al., 2021), next-generation DA tools (see Buchheit et al., 2020; Church et al., 2022; Schmidt et al., 2020) and more advanced data analytical techniques (see Eilifsen et al., 2020; Krieger et al., 2021; Spraakman et al., 2021). That said, within accounting, judging by the empirical research, DA adoption and use appears to have progressed the furthest in the area of financial auditing (for a recent review, see Ruhnke, 2023). Several studies report that the large accounting firms have begun to use commercial DA software, scripting languages and in-house tools to increase the scope and depth of their audits (Austin et al., 2021; Eilifsen et al., 2020; Kend and Nguyen, 2020; Salijeni et al., 2021). Some firms have also turned to dashboard visualization to simplify auditing work and enhance communication with clients (Salijeni et al., 2021). However, even in auditing, more advanced uses of DA seem to remain rare (see Eilifsen et al., 2020; Krieger et al., 2021). The adoption and use of DA in auditing/by auditors has been explained in different ways. Some scholars have drawn on information systems theories. Examples include the use of technology affordance theory (Salijeni et al., 2021), the technology organization and environment framework (Krieger et al., 2021) and epidemic/probit models of technology adoption (Buchheit et al., 2020). Other scholars have proposed more institutionally-oriented explanations, many of which focus on legitimacy and (failed) legitimacy building (e.g. Austin et al., 2021; DeSantis and D’Onza, 2021; Eilifsen et al., 2020). While both approaches have their merits, they do not engage sufficiently with important peculiarities of the accounting context. More specifically, they pay too little attention to the role and influence of accounting’s rich professional heritage, on the one hand, and ongoing commercialization, on the other hand. Yet, these factors constitute much of the context into which DA is introduced – in accounting. Therefore, they likely play a major role in relation to the uses and nonuses of DA in FDD and, more broadly, accounting practice. In the following, we elaborate on why this is to be expected.

Established as a “modern” profession in the 19th century in the UK (Lee, 2013), accounting – especially public accounting – has long been involved in and associated with the production and assurance of relevant and reliable financial information, information that offers a “true and fair view” of a company’s state of economic affairs. For instance, in the context of auditing, the unqualified opinion essentially resembles a seal of approval – which signals that the information is trustworthy and can be used by investors and other stakeholders for decision-making purposes. This signal is especially strong when assurance is given by one of the “elite” Big Four accounting firms (Craswell et al., 1995). Simply put, the “Big Four stamp” symbolizes trustworthiness and, in effect, gives comfort to decision-makers. More recently, as part of their commercialization (Cooper and Robson, 2006; Suddaby and Greenwood, 2005), the Big Four have leveraged this reputation to move into new areas of practice, including accounting-based advisory services. This creates an interesting situation in which the aforementioned associations carry over, as a strong selling point, but also as an expectation that has to be met – so that (the value of) the “Big Four stamp” does not erode. This risk, in turn, may constrain aggressive DA adoption and use in FDD which, otherwise, would be an excellent opportunity for developing new offerings, extending the territory of FDD work and, ultimately, generating new revenues. The emphasis thereby lies on may. As for outside expectations, clients want as much “bang for the buck” as possible. This may or may not involve expectations of extensive DA use in FDD. At the same time, clients want information that is reliable – to make decisions and valuations based on it. Similarly, capital providers and M&A insurers demand reliable information from FDD, for very similar reasons. The question is: will DA-based insights be perceived as equally reliable and trustworthy as traditional FDD output? This is unclear and introduces uncertainty in relation to DA use in FDD. To summarize, extensive integration of DA into FDD may seem enticing from a commercial perspective but may be problematic from a professional (heritage) standpoint. This means that DA use in FDD practice, likely, is influenced by commercial and professional logic. However, the nature of the influence remains to be empirically examined.

2.2 Accounting and commercial logic

The scholarly discourse on the influence of and tensions between professional and commercial logics in accounting practice builds on and mobilizes the larger institutional logics perspective (e.g. Friedland and Alford, 1991; Thornton and Ocasio, 1999; Thornton et al., 2012). In recent years, this perspective has become a popular framework to link institutions, that is, “historical accretions of past practices and understandings that set conditions on action” (Barley and Tolbert, 1997, p. 99) to organizational and individual action (Thornton and Ocasio, 2008). The basic idea is as follows: the major societal institutions (e.g. the market) each have a central logic (e.g. capitalist) that provides organizations and individuals with broad rationalizations and justifications for acting or refusing to act in certain ways (Friedland and Alford, 1991). As for a definition, as described in the introduction, following Greenwood et al. (2010, p. 521, citing Thornton, 2004, p. 70), we understand institutional logics as:

[R]esilient social prescriptions that enable actors to make sense of their situation by providing “assumptions and values, usually implicit, about how to interpret organizational reality, what constitutes appropriate behavior, and how to succeed.”

In other words, institutional logics specify what counts as meaningful and appropriate in a specific social context (Pahnke et al., 2015). As broad belief systems, they enable organizational actors to make sense of their situations and guide them in determining which behaviors to engage in (Alvehus and Hallonsten, 2022). On the one hand, they constrain, i.e. set conditions on, choice and action. On the other hand, they also offer actors opportunities to “construct and reconstruct logics in ways that reflect their interests” (Haveman and Gualtieri, 2017, p. 13). This becomes especially relevant in the context of institutional complexity, that is, when organizations or individuals “confront incompatible prescriptions from multiple institutional logics” (Greenwood et al., 2011, p. 317). Connected to that, the institutional logics perspective provides an analytical lens to examine the effects of, and the responses to, diverse institutional demands on organizations and individuals (ibid.). Below, this is explored in relation to professional accounting practice.

As implied previously, accounting firms and professionals are often tasked with reconciling two seemingly contradictory logics: professional logic and commercial logic (see Cooper and Robson, 2006; Greenwood et al., 2011; Lander et al., 2013). In that regard, according to Lander et al. (2013), changes in the institutional environment, especially technological advancements, internationalization, fiercer competition and expanding client demands, have led to a profound shift in the control of the profession (toward market and hierarchy-based controls) and, in effect, a greater emphasis on commercial values and logics. As part of that, the large accounting firms, especially the elite Big Four, started to offer more and more advisory (consulting) services in new areas of practice and, in the eyes of some, “forego their professional independence […] for the concept of value added” (ibid., p. 133). FDD is one of these advisory services; DA is meant to be a value-add (see Bose et al., 2023). At the same time, as noted before, the professionalism that is ascribed to the elite accounting firms continues to be a key selling point for their advisory services – including FDD. So far, and in the above, commercialism and professionalism have been presented as stark contrasts. However, after many decades of managing and blending these logics, this dichotomy might have become misleading. Today, honoring the commercial aspect of professional accounting work is basically part of being a professional (see Anderson-Gough et al., 2022; Faulconbridge and Muzio, 2016). That said, accounting still poses many technical questions, which means that acquiring and applying technical expertise is also part of being an accounting professional (see Spence and Carter, 2014). For this reason, it appears preferable to conceptualize professionalism as a broader logic, one that incorporates seemingly contradictory value clusters (ibid.; Suddaby et al., 2009), and distinguish between technical-professional logic and commercial-professional logic (Spence and Carter, 2014). For convenience, we use the terms accounting logic and commercial logic hereafter. Drawing on Greenwood et al. (2010, p. 521) and Spence and Carter (2014, p. 949), we define and understand accounting logic (commercial logic) as the:

Resilient social prescriptions that enable actors to make sense of their situation by providing assumptions and values about how to interpret organizational reality, and what constitutes appropriate behavior, regarding the issues that are embodied in the delivery of technical accounting work (in relation to the commercial aspect of accounting practice).

To make both logics more concrete, below, we give a short overview of central elements that have previously been ascribed to them.

Accounting logic includes an implicit assumption that organizational activity can and should be evaluated in terms of visible and verifiable financial in- and outflows (Broadbent 1998, 2002) and, in effect, prioritizes financial aspects and understanding. It also encompasses certain ideas and beliefs about (the importance of) auditability and traceability (see “audit trail”), checks and balances, prudence and conservatism and reliability and accuracy (Power 1997, 2005) – ideas and beliefs that mold accounting professionals’ understanding of “how accounting should be performed and what it should look like” (Heinzelmann, 2017, p. 165). Commercial logic, on the other hand, includes an orientation toward revenue generation and maximization (Brivot et al., 2015; Spence and Carter, 2014), profits and profitability (Cooper and Robson, 2006; Malsch and Gendron, 2013) and billable hours and engagement fees (Brivot et al., 2015). Moreover, it entails a concern for expansion and business opportunities (Dunne et al., 2023), economic value and added value (Järvenpää, 2007; Spence and Carter, 2014) and efficiency and rationalization (Brivot et al., 2015; Spence and Carter, 2014). Finally, commercial logic foregrounds serving client interests, needs and relations (Dunne et al., 2023). To conclude, we wish to reiterate the usefulness of the distinction between accounting logic and commercial logic – in relation to DA use in FDD practice and, more broadly, accounting practice. DA is poised to have a significant impact on the delivery of technical accounting work and the commercial aspects of accounting practice (see 2.1). In other words, DA use will affect aspects of accounting for which accounting logic and commercial logic provide resilient prescriptions. Thus, a strong link between the three seems highly probable. Furthermore, exploring the influence of accounting logic on DA use, in light of commercial logic, could lead to insights into how accounting logic impedes or, perhaps, is implicated in DA-based value creation. Conversely, examining the influence of commercial logic, in light of accounting logic, could lead to insights into DA’s business value in accounting. These relations motivate the following research question:

RQ.

How do accounting and commercial logic influence the use of data analytics tools in financial due diligence?

3. Materials and method

3.1 Context and case

In M&A, buyers have an information disadvantage. To reduce this disadvantage, they perform or commission what is called due diligence (Lajoux and Elson, 2011). For the financial part, i.e. FDD, they typically engage one of the large accounting firms. In most cases, FDD involves two major, interrelated areas: historical performance and accounting adjustments. The former is about understanding what has driven profits, costs, etc., and the latter is about identifying accounting adjustments, primarily related to underlying earnings, net debt and net working capital (Pomp, 2015). To perform these analyses, FDD advisors often rely on generic financial analysis techniques, such as horizontal, vertical or ratio analyses. In addition, they may apply industry-specific analyses, such as price-volume-mix analyses (in retail). Traditionally, FDDs have been based on targets’ internal data, i.e. financial statements, management accounts and relevant supporting documentation. This encompasses (monthly) management reports, usually with detailed financial and operational information on different parts of the business, including key performance indicators. Notwithstanding these customs, in principle, any data, internal or external (provided it is made available), may be used. This includes nontraditional and/or external data sources, such as socioeconomic information, weather data or geo data.

Whereas empirical research on DA use in accounting is scarce, research on DA use in FDD is almost nonexistent (Neumann’s (2020) doctoral dissertation being a noteworthy exception). We, therefore, conducted an exploratory case study of the FDD practice of a Big Four accounting firm in Sweden (pseudonym: DealCo) (see Edmondson and McManus, 2007). For more than a decade, Excel had been the industry standard for data analysis. Lately, however, the Excel-only approach had reached its limits because data sets had grown too large for (efficient) processing. In response, DealCo’s management concluded that they needed to innovate their FDD practice. As part of this initiative, proprietary in-house DA tools, including SalesAnalyzer (for processing and analyzing transaction-level data) and AutoConverter (for converting SIE files – standard import and export: an open standard for accounting data in Sweden – into structured Excel and PowerBi outputs), were developed, licenses for commercial DA tools, including Alteryx (a low-code/no-code end-to-end DA platform that enables analytic process automation) were acquired, junior staff were given training on using the tools and an informal Data and Analytics (D&A) group was established at DealCo’s FDD department. For the most part, these efforts were driven by accounting professionals. When the first interviews were conducted, the D&A initiative had been ongoing for about a year.

3.2 Data collection

The present study forms part of a larger research project on digital transformation in FDD (Kastrup et al., 2024) [1]. The primary data (see Table 1 for an overview), most of which was collected between October 2021 and June 2022, comprises a kick-off meeting, 13 semistructured interviews, two walkthrough sessions, a close-out meeting and concluded with a short one-year follow-up in May 2023 (see “prolonged engagement,” Lincoln and Guba, 1985) which allowed us to confirm that the paper’s findings were still current at that time. Interviewees included managers as the key informants, but also partners/directors, and (senior) analysts/associates. Contact was established through the manager (M4) who co-led DealCo’s D&A group. An interview guide was used that centered around DA, data and judgment. The data used in this paper stems from the sections on DA and data. While keeping with these themes, questions were modified so they would match the experience of the interviewee. The interviews, all conducted in English, were recorded and transcribed. Since the interviewees were highly proficient in English, we did not experience language issues. In the two walkthrough sessions, M4 demonstrated on screen how they had used different DA tools in previous projects.

Because of COVID-19, all interviews and walkthrough sessions were conducted via Zoom. While recognizing potential downsides of video interviewing (e.g. picking up on nonverbal cues), in this case, the interviewees seemed very experienced in and comfortable with meeting online. This contributed to having open, high-quality interviews. During the kickoff meeting, DealCo’s attendees summarized their DA journey (until that point), highlighted the most pressing challenges and outlined their DA ambitions for the future. During the close-out meeting, we were able to confirm that our analyses and conclusions were in line with how DealCo’s attendees experienced and perceived DA and DA use (in addition, a later draft of the manuscript was sent to DealCo for review and “member check,” see Lincoln and Guba, 1985). The kick-off meeting, close-out meeting and walkthrough sessions were not recorded; notes were taken. Interviews with multiple advisors from each rank, corroborated by observational data in the form of two walkthrough sessions, on the one hand, and close-out meeting, one-year follow-up and the abovementioned member-check, on the other hand, enhance the credibility of the reported findings (see Lincoln and Guba, 1985). Additionally, different types of secondary data were collected. This includes company videos, podcasts, presentations, a Webcast, website posts and white papers – created by DealCo’s global network of member firms. These materials were publicly available (i.e. openly accessible online). Taken together, these materials helped us to gain an overview of new data, new DA tools and new DA applications. This knowledge, in turn, greatly aided in conducting and making sense of the interviews. Table 2 gives an overview of the collected secondary data.

3.3 Data analysis

The data analysis was inspired by Braun and Clarke’s (2006) thematic analysis and included inductive and deductive coding elements (Fereday and Muir-Cochrane, 2006). The first round of coding was guided by the empirical question of how DA tools were used in FDD and the desire to explain why they were used this way. Apart from “DA tools” and “use,” the first round did not involve any predetermined concepts or codes. This produced dozens of first-order codes, many of which seemed to pertain to norms and practices of data and model use and a highly fragmented explanation for DA use in FDD. In search of a less fragmented explanation, we then compared these codes against some more general concepts and theories. This led us to conclude that the aforementioned norms and practices were perhaps best thought of as building blocks and manifestations of different institutional logics. Following this line of thought, we adopted the institutional logics perspective as a theoretical lens, respectively, method theory (Lukka and Vinnari, 2014). Hence, in hindsight, with respect to capturing institutional logics, the first stage involved a “pattern inducing” process as described by Reay and Jones (2016). As an analytical technique, pattern inducing is credited for capturing and foregrounding the nuances of localized practices and actors’ explanations of values and beliefs (ibid., p. 443, Table 1). This, generally, leads to accurate accounts that, however, may not generalize – in all aspects – beyond the local context (ibid.). In Section 6, we expand on this point as we discuss transferability. When reasoning around their (non)use of DA tools, the interviewees often referred to accounting traditions and commercial considerations. Based on their descriptions in the extant literature, we then adopted accounting logic and commercial logic as key constructs (see 2.2). Moreover, we noticed that how the advisors thought and talked about DA differed markedly from how DA was described in much of the literature (see Davenport and Harris, 2007; Delen and Zolbanin, 2018; Nielsen, 2018). This led us to conclude that the advisors entertained an altered DA logic. To that end, it should be noted that when talking about DealCo’s DA logic, “logic” refers to the more general meaning of the term, i.e. a particular way of thinking about something – DA in this case. Based on these concepts (accounting logic, commercial logic and DA logic), we performed a second round of coding. This round or phase combined both “pattern inducing” and “pattern matching” (see Reay and Jones, 2016). The latter, i.e. pattern matching, enables comparison to other studies; on the flipside, mobilization of established theory might limit the emergence of new insights (ibid.). In this study, starting with pattern inducing largely curbs this issue. The pattern-inducing part was about identifying those elements of accounting logic and commercial logic that seemed to influence DA use in FDD most strongly. To capture these, we examined the symbols and beliefs expressed in discourse, i.e. during the interviews (e.g. “it has to be auditable”), and material practices that could clearly be associated with an institutional logic (e.g. reconciliations with accounting logic) (see Reay and Jones, 2016). Following Braun and Clarke’s (2006) recommendations, relevant passages were thus coded into first-order codes which, after some iterations, combined into several second-order themes (see the subthemes in Figures 1 and 2). In a final step, these (sub)themes were combined into two overarching themes: accounting/commercial logic as a driver of DA use, and accounting/commercial logic as a hindrance of DA use. Accounting and commercial logic as drivers and hindrances gives a voice to different constructions of reality which, according to Lincoln and Guba (1985), is a mark of credibility in qualitative research. Importantly, Figures 1 and 2 are not meant to be exhaustive. Rather, they show the most salient links between accounting logic, commercial logic and DA use in DealCo’s FDD practice. The pattern-matching part was about comparing these logics with their descriptions in the extant literature. This was done “to paint a likeness of” (see Reay and Jones, 2016, p. 442) these logics – as they manifested in the FDD context – in the light of their prior (i.e. non-FDD specific) characterizations. In addition, to encapsulate the essence, or the cornerstones, of DealCo’s DA logic, we coded for two types of accounts: thinking/reasoning about DA (i.e. symbols and beliefs expressed in discourse) and actual use of different DA tools (i.e. material practices associated with the altered DA logic). This assumes that logics manifest in behavior/action (see Haveman and Gualtieri, 2017). Relevant accounts were then coded into first-order subthemes which, after iterations, combined into three second-order themes, namely, depth, efficiency and effectiveness (see Figure 3). The breadth of findings necessitates an economical presentation of the empirical materials. Nevertheless, Figures 1, 2 and 3, which depict themes, subthemes and illustrative quotes, make it possible to trace the central conclusions back to the underlying data (see “confirmability,” Lincoln and Guba, 1985).

4. Findings

In the following, we present the findings from our study of DA use in DealCo’s FDD practice. First, we detail the most salient links between accounting logic and DA use (4.1). Thereafter, we detail the most salient links between commercial logic and DA use (4.2). Finally, based on 4.1 and 4.2, we highlight a less salient but equally important cognitive link between accounting and commercial logic and DA use in DealCo’s FDD practice (4.3).

4.1 Accounting logic: Hindrance and driver

Whenever we get a data source, we always have to reconcile it. We call it audit trail. We always go to the audited financials and then have a trail to wherever that data is. Actually, we spend a lot of time trying to reconcile the data dump that we have to the management accounts. To understand, to make sure that we can use the data to make analyses. Because the worst thing we could do is spend a week analyzing it and then realizing it does not tie to anything, so it’s completely meaningless […]. Does it tie in? Plus or minus five percent. Does it make sense in the financial scheme? (P1)

As presumed, in DealCo’s FDD practice, accounting logic conflicted with DA use in several ways. Tensions were most salient in relation to the advisors’ rigid financial focus, their fixation on traceability and their concerns about data integrity (see Figure 1). Below, we explore each of these themes.

When discussing the possibilities for integrating alternative data, e.g. external nonfinancial data, into FDD practice, almost all the interviewees noted often unclear financial implications. This was evident in questions such as “but does it link to the financials?”, “does it make sense in the financial scheme?” or “what does it tell us about the financials?” This matches closely with Broadbent’s (1998, 2002) characterization of accounting logic as centering around visible and verifiable financial in and outflows (see financial focus). Closely related, tensions between alternative big data and accounting logic were also evident in relation to the accounting practice of reconciling data with an authoritative second account, preferably, the audited financials. If the data reconciled, they were considered trustworthy and meaningful. In discourse, this was evident in expressions such as “we always have to reconcile it” or “if it reconciles, we can use it to make analyses.” Yet, most external data could not be reconciled this way, which made the advisors hesitant toward using them. Moreover, against the backdrop of intensive time pressure in FDD, DealCo’s advisors were wary not to waste valuable hours or days analyzing data that, in the end, “did not tie” into the financial scheme of the business.

Tensions also became salient when discussing the potential for more sophisticated forms of modeling, e.g. predictive DA, in FDD. DealCo’s advisors all stressed the importance of audit trails and working in ways that made it easy for others to trace conclusions all the way back to the source data (hence “traceability”). For instance, some noted that analyses and conclusions needed to be “fully transparent” whereas others cautioned that “you cannot have a black box.” In large parts, this was linked to the larger context in which FDD is situated. FDD deliverables – analyses, conclusions and reports – inform different M&A downstream activities (valuation, pricing negotiations and financing) and thus have to withstand the scrutiny of clients, targets and capital providers. For this reason, DealCo’s advisors were highly skeptical of complex DA solutions where their understanding of the underlying process would have been rather limited and traceability would have been comparably poor. The concern for traceability, or auditability, closely matches Power’s (2005) characterization of accounting logic, especially the desire “to make things auditable” (see Power, 1997).

More tensions between accounting logic and DA became evident in relation to (perceived) data integrity. Indeed, low or lacking data integrity (“the accuracy, completeness, and quality of data as it’s maintained over time and across formats,” Cote, 2021) was repeatedly mentioned as a key reason for not incorporating external data more often. In line with accounting logic, the targets’ accounting data, especially when audited, were seen as comparably accurate and reliable while many alternative big data sources (e.g. social media) were seen as comparably noisy and unreliable. Whereas the former was viewed as something “that actually you know what it is” the latter was, somewhat contemptuously, referred to as something “we find on the internet” and that, perhaps, could not be fully trusted. Among DealCo’s advisors, there was, therefore, little desire to “mix” different types of data and contaminate accurate and well-understood (accounting) data with noisy and less well-understood (external) data. These views and preferences reflect the verification aspect of accounting logic noted by both Power (1997) and Broadbent (2002).

Notwithstanding the abovementioned tensions, in DealCo’s FDD practice, accounting logic also drove DA use in several ways. This was most salient in relation to the prospects of better financial understanding, the possibility of more granular analyses and the creation of trust and comfort. Below, we explore these themes.

For a long time, (generic) financial analysis techniques, which we view as manifestations of accounting logic, had been trusted means for assessing the state and the prospects of a business and explaining its financial whats, hows and whys: horizontal analyses reveal trends and patterns; vertical analyses show how numbers build up and explain changes between periods. Consistent with these practices, the advisors’ work revolved around creating understanding (of what was driving the business) and giving explanations (for why things were the way they were). Indeed, many advisors frequently invoked the terms “understanding” and “explanation” when talking about the overarching objectives of their efforts. This suggests that everything that contributed to a better financial understanding of the target’s business was, in principle at least, meaningful in FDD. For instance, M4 recalled a previous project in which they incorporated weather data to make sense of the financial performance of a heating-related business and “understand what made that company tick.” Similarly, P1 referred to an example of a foreign currency exchange platform where using DA had made “hundreds of millions of transactions very understandable.”

Often, though not always (see the weather data example), better financial understanding was achieved by performing trusted industry-specific analysis, e.g. price-volume analyses (retail), customer churn analyses (software) and capacity analyses (construction), with highly-detailed data. Consider the example of price-volume-mix analyses (or “sales bridge”). For these, newly available data, e.g. sales data on a receipt basis, made it possible to reach an unprecedented level of depth, that is, perform more granular analyses than before. Higher resolution contributed to establishing, with more confidence, if certain revenue or margin developments were driven by unit sales, better marketing, smarter pricing or systemic changes in the market. This, however, could only be done by leveraging the data-processing capabilities of next-generation DA tools like Alteryx and SalesAnalyzer. It is worth noting that much of this change was about the depth rather than the type of analysis. Price-volume-mix analyses have delivered financial intelligence for a long time. Yet, with transaction-level data, they delivered deeper financial understanding and more reliable financial intelligence.

In fact, the use of transaction-level data and, by implication, DA tools was supported by yet another core accounting practice: reconciliations. This is interesting as it shows that accounting practices – even one and the same accounting practice – might hinder and drive DA use. Unlike many other big data sources, transactional data were, in principle at least, reconcilable with the latest (audited) financial report. And, as mentioned before, this exercise – reconciling new data – was often central for DealCo’s advisors in determining whether data sources were trustworthy and worth analyzing. If data reconciled, they were viewed as “safe to use” and the advisors felt comfortable about slicing and dicing the data ad libitum using new DA tools. This fits closely with the idea of accounting as a trust and comfort-producing activity (Pentland, 1993; Porter, 1995; Power, 1997). In summary, in DealCo’s FDD practice, accounting logic and associated accounting practices, both hindered and drove the use of DA.

4.2 Commercial logic: driver and hindrance

In terms of the value of doing stuff [i.e., DA], in the end, it comes down to our clients as well. I mean, it’s obviously a commercial business. We want to deliver as much value as we can and like to go into detail in the analysis of complex data. But we are not going to go into the details if there is not something that generates value for our clients. (M1)

As expected, in DealCo’s FDD practice, commercial logic drove DA use in several ways. This was particularly evident in relation to the prospects of time savings, higher product quality and (sustaining) competitive capacity (see Figure 2). Below, we explore each theme.

FDD is a consulting business with hourly billing. Since the clients are buying their advisors’ time, to stay relevant in the market, it is imperative not to waste time and to work as efficiently as possible. Against this background, it is not surprising that many of DealCo’s advisors stated time savings as the biggest benefit of using DA tools. This was evident in expressions such as “it’s a time saver,” “it helps us reach conclusions faster” or “get to the 80–20 stuff [quicker].” This was especially apparent in relation to preparing data and building databooks. In the larger deals, before the introduction of new DA tools, it had sometimes taken them weeks to compile all the data and put them in neatly structured databooks. With Alteryx, and when leveraging its automation capabilities, DealCo’s advisors performed the same tasks in days rather than weeks. Some estimated that using Alteryx instead of Excel had shortened data preparation times by up to 80%. The above reflects the efficiency and rationalization aspect of commercial logic (see Brivot et al., 2015; Spence and Carter, 2014) and illustrates how commercial logic drove DA use in FDD practice.

Connected to that, one should, however, note that the “obsession” with efficiency was not necessarily about finishing projects faster and doing more projects. Rather, the main goal was to spend as little time as possible on low-value-add tasks (e.g. data preparation) and as much as possible on high-value-add tasks (e.g. data analysis). To this end, the reasoning was simple: save time in the data preparation, use this time for additional/deeper analyses, generate more decision- and value-relevant insights and, ultimately, deliver a higher product quality. Indeed, in this context, the advisors frequently invoked commercial language and terms, such as “the product” (i.e. the due diligence report), “product quality” (i.e. report quality) and “product leader” (i.e. the engagement manager). Moreover, “added value” was an omnipresent point of reference in the advisors’ discourse. This matches prior descriptions of commercial logic in the accounting literature (see Järvenpää, 2007; Spence and Carter, 2014).

It also implies a direct link between the use of DA tools and DealCo’s competitive capacity (in the FDD market). The aforementioned DA opportunities were available and known to all its competitors. Because of this, a failure to exploit these opportunities would have, in the mid to long run, harmed DealCo’s ability to compete, especially when tendering for larger, more data-intensive, high-revenue projects. Most of DealCo’s DA efforts were directed at making internal work processes more efficient – they pertained to the efficiency dimension of competitiveness. However, as mentioned before, processing data more efficiently also made it possible to widen and increase the scope and depth of analysis in FDD and, in effect, deliver higher product quality. Thus, indirectly, DealCo’s efficiency focus led to greater effectiveness (understood as doing the right things that bring into effect desired outcomes), too. In part, this relates to the business opportunities and expansion dimension of commercial logic (see Dunne et al., 2023; Spence and Carter, 2014).

Notwithstanding the above, somewhat surprisingly, in DealCo’s FDD practice, commercial logic also conflicted with DA use in several ways. Tensions were especially salient in concerns and considerations regarding chargeable hours, DealCo’s value proposition and its competitive strategy. Below, we explore these themes.

Working efficiently was important to be and stay competitive in the FDD market. However, higher efficiency could also mean fewer chargeable hours and, ultimately, less revenue. One analyst/associate remembered a project with many entities, i.e. with lots of data wrangling, in which the old Excel-based approach would have generated millions (SEK) of revenue. Using the more efficient Alteryx-based approach, they “didn’t even get close to it.” Indeed, “getting paid” and “clients’ willingness to pay” were frequently brought up when discussing potential hindrances of DA use in FDD with the advisors. This illustrates a marked concern for revenue generation and maximization, as is associated with commercial logic (see Brivot et al., 2015; Spence and Carter, 2014). Also, few of DealCo’s clients were fully aware of the possibilities that DA had afforded to FDD practice. Getting client-facing DA uses (e.g. SalesAnalyzer in retail cases) in the scope of work, therefore, often required considerable persuasion and skillful selling. While the possibilities for interesting analyses were endless, there were not quite as many that clients were willing to pay for. In consequence, client-facing DA use, beyond the agreed-upon scope, remained a risky business with an uncertain pay-off.

Interestingly, skepticism toward the value or importance of DA in FDD existed not only on the clients’ part but also on DealCo’s side. Some of the more seasoned advisors believed that the ability to perform DA was “not a key selling factor” and “not a differentiator,” and thus not central to the value proposition. Rather, it was thought of as something DealCo had to lean into more if, and only if, its competitors did. What is more, given the tight timelines, DealCo’s advisors tried to apply Pareto’s 80/20 rule in their work, i.e. to focus on the vital few, that is, the key drivers of the business. If something was unlikely to influence the client’s eventual decision and did not have clear implications for the valuation or pricing of the business, then, it probably was not worth looking into. And, indeed, there was the suspicion that many DA-based analyses would produce these types of insights: interesting, but without immediate decision- or value-relevance. Also, given the high time pressure in FDD, collecting more data to perform additional analyses to validate what was already assumed, e.g. using external big data to produce corroborating evidence, had limited appeal.

In addition, DealCo’s advisors hinted at tensions between DA and parts of their competitive strategy. Among them, DA was strongly associated with standardization – which was key to scaling DA solutions and making them profitable. The problem was: DealCo tailormade a lot of its work to meet specific client needs; customization was an important part of its competitive strategy. Also, DealCo’s advisors took pride in how accommodating they were to their client’s needs – something they believed made them stand out from the competition. On the one hand, this reflects the (serving) client interests, needs and relationship aspect ascribed to commercial logic (Dunne et al., 2023). On the other hand, it relates to competitive strategy. This aspect has been absent from prior characterizations of commercial logic. However, given that competitive strategy is inseparable from future value and revenue generation (see Porter, 1980), we contest that the business strategy aspect could hardly be more central to commercial logic. In summary, commercial logic, like accounting logic, drove and hindered DA use in FDD practice.

4.3 Accounting logic and commercial logic: more than the sum of their parts

The one we use a lot is Alteryx. This is a tool that is very well suited if you have a lot of data. In Excel, there are only a million lines or so [laughter]. Whereas in Alteryx, I processed maybe 20 million data points, etc., so it enables us to work with larger data sets. That is one of the reasons we use it. Then, when I think about different applications, it is a lot about data processing – making it more efficient when there is a lot of data. […] I wouldn’t say that we apply any sort of machine learning or that the tools make the decisions for us. It’s more about data processing. (M4)

The preceding sections have highlighted the most salient links between accounting logic (4.1), commercial logic (4.2) and DA use in DealCo’s FDD practice. Adding to that, in what follows, we draw attention to a less salient (indirect) but perhaps more general mechanism of influence, namely: logics as perceptual and conceptual filters (see Bettis and Prahalad, 1995; Schraven et al., 2015; Von Krogh et al., 2000).

As mentioned previously, DA is most commonly understood as the “extensive use of data, statistical and quantitative analysis, explanatory and predictive models […] to drive decisions and actions” (Davenport and Harris, 2007, p. 7). That, however, was not how DealCo’s advisors understood DA. For instance, some pointed out that “if the definition was insights from data,” it was “to some extent decoupled” from (some of) the tools they were using. Others thought of DA as “a tool to save time” rather than “to help with decision making.” Indeed, among DealCo’s advisors, DA was primarily seen as a means to handle big data, save time and get raw data into a structured shape. These perceptions, in turn, seem to reflect some of the subthemes associated with accounting logic (e.g. better financial understanding through more granular analyses) and commercial logic (e.g. better product quality through time savings) identified in Sections 4.1 and 4.2. This implies that accounting logic and commercial logic, to some extent, shaped the advisors’ perceptions of what DA was “good for” in FDD and, in effect, their conceptual understanding of it. The former was perhaps most visible in DealCo’s advisors’ use of Alteryx (a commercial end-to-end DA platform) and the purpose of SalesAnalyzer and AutoConverter (the two proprietary tools): Alteryx was mainly used to process and structure data faster and perform more granular analyses; SalesAnalyzer was specifically developed to do these exact things with transactional data; and AutoConverter was specifically developed to speed up the processing and structuring of SIE files (an open standard for accounting data in Sweden).

This perceptual focus on efficiency-oriented DA use cases, in turn, appeared to give rise to efficiency-centric conceptual understandings of DA. Regarding that, it is worth noting a subtle difference in individual DA conceptions. Some of DealCo’s advisors viewed DA merely as an “efficiency tool” and others viewed it as “a means to handle big data efficiently.” Presumably, this difference reflects differences in relation to first-hand experience with using DA tools to process big data, which some had more than others. That said, even the former, i.e. the more experienced, associated DA with big data and efficient data processing. For one thing, these understandings differ markedly from the “novel insights, better decisions” conceptualizations of DA found in much of the literature (see Davenport and Harris, 2007). For another thing, they were part of a specific DA logic, henceforth referred to as “DealCo’s DA logic,” which seemed to be the nexus between institutional influence and DA use in DealCo’s FDD practice. Below, this logic, and its components, are analyzed.

As shown in the preceding sections, in DealCo FDD practice, DA and DA tools were mostly thought of and used as a means to perform more granular analyses, process data more efficiently and achieve better focus in the analysis. This points toward a specific DA logic that revolved around depth, efficiency and effectiveness (see Figure 3). Although the depth aspect was important, the efficiency and effectiveness aspects constituted the core of DealCo’s DA logic: by ways of standardizing and automating, it was possible to save time on data preparation and spend more time on the actual analysis (see Figure 3, shaded area). What is important to point out, the focus on saving time was a matter of efficiency (“doing it right”) and effectiveness (“doing the right thing”): because of the time pressure, operational efficiency was needed to ensure that there was sufficient time for doing the right thing, that is, doing the analyses and drawing the conclusions that were most valuable to DealCo’s clients. A3’s quote below captures the essence of this logic:

Why we focus so much on being efficient and saving time is just that, by doing that, we can also spend more time analyzing it and we get a higher quality of our report and of our product.

In line with this logic, each of DealCo’s DA tools, in one way or the other, contributed to greater efficiency and effectiveness: Alteryx was used for a wide range of data processing and structuring tasks; SalesAnalyzer generated a standardized reporting package and a data cube for customized analyses; and AutoConverter automated the conversion of SIE files into neatly structured Excel and PowerBI outputs, which then aided the identification of focus areas for the analysis.

The three components described above (depth, efficiency and effectiveness) point toward a blending of accounting and commercial logic in DealCo’s DA logic. This is evident in what is part of the logic and what is not. Commercial logic – which postulates that efficiency and effectiveness, i.e. “doing it right” and “doing the right thing,” are focal to creating added value, producing a competitive offering and ensuring commercial success – is present in the concern for and focus on these ideals (i.e. efficiency and effectiveness). Moreover, it seems present in the absence of concern for and focus on more sophisticated forms of insight-oriented DA that were not perceived as key value-adds (see 4.2). Accounting logic – which assumes that visible and verifiable financial in and outflows (and, thus, transactional data) are critical for evaluating and understanding organizational activity – is present in the depth component (which connects mostly to the analysis of granular transactional data). Moreover, it seems present in the absence of components involving external data and complex models which were deemed as problematic from a data integrity and auditability standpoint (see 4.1). In other words, the evidence implies that DealCo’s advisors viewed DA – both in terms of possibilities and value – through the lens of accounting logic and commercial logic. As perceptual and conceptual filters, they seemed to draw attention to some action possibilities (e.g. using DA for greater efficiency) and divert it from other action possibilities (e.g. using DA for predictive modeling). Regarding the initial question of how accounting and commercial logic influence the use of (new) DA tools in FDD, this implies a second, more indirect link: as perceptual and conceptual filters, accounting and commercial logic seemed to shape the advisors’ understanding of DA and their assessment of what it was good for in FDD – as reflected in DealCo’s DA logic – and, through that, their use of new DA tools. The significance of this is discussed in 5.3

Finally, Figure 4 visually summarizes how accounting and commercial logic influenced the use of new DA tools in DealCo’s FDD practice and, thereby, provides a concluding answer to the research question. The solid arrows represent the direct link: accounting and commercial logic as drivers of and hindrances to DA use (see Figures 1 and 2). The dotted arrows represent the indirect link: accounting and commercial logic, acting as perceptual and conceptual filters, shaping DealCo’s DA logic (see Figure 3) and, in effect, the advisors’ use of new DA tools. In particular, Figure 4 foregrounds the link between abstract conceptions of DA and what it is good for (DealCo’s DA logic), on the one hand, and the concrete use of new DA tools for deeper analyses, faster processing and better focus in FDD, on the other hand. At large, it points toward DealCo’s DA logic as the nexus between institutional influence and DA use in its FDD practice.

5. Discussion

In the following, we discuss our findings and highlight how they contribute to the literature on digitalization in accounting – especially in relation to DA and accounting for decision-making. The discussion is structured along three themes: accounting-analytics tensions (5.1), analytics’ business value (5.2) and conceptions of analytics (5.3).

5.1 Accounting-analytics tensions

Prior field research on big data and accounting pointed toward tensions between the traditional “accounting mindset,” evident in the reports of “reluctant accountants” with a “rigid focus” on traditional data and measures, and the new DA approach (e.g. Agostino and Sidorova, 2017; Arnaboldi et al., 2017). In contrast, our study suggests that it may often not be as clear-cut as previously assumed and reported.

In DealCo’s FDD practice, accounting logic, the institutionalized prescriptions regarding the technical delivery of accounting work, conflicted with DA in numerous ways. The most salient tensions related to the focus on financials, concerns around traceability and beliefs about data integrity. As one of DealCo’s managers put it: “Traditionally, the FDD has been based more on historical observable data […] always based on reported audited numbers that are verifiable in a sense.” Implicit in this statement is the assumption that organizational activity can and should be evaluated in terms of visible and verifiable financial in- and outflows which is foundational to accounting logic (see Broadbent, 1998). Furthermore, perceived traceability appeared to be a particularly difficult-to-jump hurdle for more advanced forms of DA use. In that regard, the advisors noted repeatedly how important it was that they could explain what they saw (i.e. why are the margins for product A so high in market B?) and what they had done (i.e. how did you transform input C into output D?) and that third parties could easily trace back conclusions to the source data. These remarks or concerns relate to the broader ideal of auditability, which is another cornerstone of accounting logic (see Power, 1997). Connected to that, since traceability was considered imperative (“you cannot have a black box,” see 4.1), many of DealCo’s advisors were unsure if there was a case for more advanced forms of DA and its sometimes opaque models in FDD. That, however, was only one side of the story. The other, much less discussed side of the story is one in which accounting logic drives DA use. For instance, transaction-level data (e.g. sales data on a receipt basis) made it possible to perform “traditional” accounting analyses (e.g. price-volume-mix analyses) on a far more granular level and, in effect, provide more reliable financial intelligence (see Bhimani and Willcocks, 2014). Closely connected to that, another core accounting practice, reconciliations, helped to establish trust in these data. This suggests that there is no inherent tension between accounting logic and big data. Rather, tensions seem to arise when data are difficult to map into the financial scheme of a business and lack clear financial implications (e.g. social media data).

These insights not only qualify prior assumptions about accounting professionals’ reluctance toward using certain types of big data but they also suggest that accounting logic could play a vital part in DA-based value creation in accounting practice. What is too often forgotten, big data and DA are not inherently valuable. They are valuable to the extent to which they inform managerial decision-making and action at a reasonable cost. Also, some big data sources include decision-critical information while others do not. Indeed, given the recent explosion of data sources, the ability to separate the wheat, i.e. reliable and relevant data, from the chaff seems to be more important than ever. In light of this, following accounting logic could help in making sure that efforts and resources are directed at analyzing and drawing conclusions from high-quality data with clear financial implications. In this scenario, accounting logic would be a guidepost, rather than a hindrance, to DA-based value creation in accounting practice. This questions the idea that to realize the value of DA accounting professionals must do away with old accounting logics and adopt the new DA mindset (see Mintchik et al., 2021; Richardson and Watson, 2021; Schmidt et al., 2020). This possibility, i.e. accounting logic as a guidepost to DA value creation, creates an alternative viewpoint and narrative in a discussion that seems dominated by accounts that, for the most part, appear to classify accounting logic as a problem and hindrance. More broadly, it links to the question of DA’s business value in accounting – which is discussed next.

5.2 Analytics’ business value

In the prior literature on digitalization in accounting, embracing DA has often been presented as an economic imperative (see Brands and Holtzblatt, 2015; Nielsen, 2018; Schmidt et al., 2020) and the assumption that DA is a value-add has often remained unquestioned. Our study and, more specifically, its findings regarding commercial logic’s influence on DA use in FDD, in part, challenge and qualify these ideas.

In DealCo’s case, commercial logic was driving DA use in some ways, in particular, through prospects such as saving time, improving product quality and sustaining competitive capacity (see Figure 2). However, commercial logic also provided numerous rationales for not leaning into DA more strongly. Some were specific to DealCo (e.g. standardization vs customization), others were specific to advisory (e.g. time savings vs billable hours) and yet others were more general. This is interesting, not least because it highlights conflicting prescriptions within commercial logic rather than between accounting logic and commercial logic – which is in line with more recent research on logic multiplicity in accounting firms and somewhat harmonious blending of these logics (e.g. Anderson-Gough et al., 2022). While accounting logic involved rather clear prescriptions on what to (or not) use DA for in FDD, commercial logic was much more ambiguous and, often, only conclusive after further specification of key objectives (e.g. is the objective to maximize present or future revenue generation?). This hints at the existence of contradictory value clusters – such as customer value and firm value or short-term value and long-term value – within commercial logic (see Suddaby et al., 2009 – who made this point for professionalism, see 2.2). Our study, therefore, cautions against viewing commercial logic as a homogenous construct. Also, coming back to commercial logic’s rationales for not leaning into DA more strongly, DealCo’s advisors were generally unsure that doing more “DA stuff” would necessarily benefit their clients, as the analyses with the clearest value implications, i.e. with a direct impact on valuation or pricing (e.g. underlying earnings, net debt and net working capital) were typically a matter of traditional accounting work rather than DA wizardry. These findings support the idea that the value of DA in accounting is "determined by the decision context and tasks at hand” (Rikhardsson and Yigitbasioglu, 2018, p. 42).

In part, this links back to accounting’s heritage, more specifically, its role in and reputation for the provision and assurance of relevant and reliable (financial) information (see 2.1). Given the nature of the exchange, DealCo’s clients could have asked for all the advanced DA – i.e. predictive and prescriptive analytics (Delen and Ram, 2018) – in the world. But they were not (yet). Instead, they were looking and paying, for accurate descriptions, reliable diagnoses and conclusive explanations. In other words, they were looking for affirmation, a seal of approval, that the targeted investment was financially sound. This suggests that, in accounting, especially in the context of high stakes decision-making (like M&A), the real value of DA may not always lie in (more complex) forward-looking uses, as is often proclaimed (see Tschakert et al., 2016; Nielsen, 2018). As for why, more complex forms of DA may often not be that suitable for the task at hand which, in accounting and auditing practice, often entails the production of comfort (see Carrington and Catasús, 2007; Pentland, 1993). For instance, FDD is meant and tasked to give M&A decision-makers comfort in, among other things, the firm’s fundamentals and trends relevant to the investment thesis. This comfort, in turn, is often key for deals to go through. For now, it is difficult to imagine how somewhat opaque predictive algorithms that might mobilize new, comparably poorly understood data will give M&A decision-makers the same comfort as simpler, more transparent analytical techniques that use traditional accounting data (on that, see Fahlevi et al., 2022, and their reason for building simple regression models in a clinical decision-making context). In light of this, i.e. considering what accounting (information) is often valued for, it becomes less self-evident that extensive and advanced DA use is an economic imperative and, thus, an urgent must in accounting (“embrace or face extinction,” see Schmidt et al., 2020). This adds a much-needed counter perspective to prior discussions around DA’s business value in accounting and has important implications for practice – which are addressed in Section 6.

5.3 Conceptions of analytics

Finally, this study extends prior institutional theorizing of DA adoption and use in accounting (e.g. Austin et al., 2021; DeSantis and D’Onza, 2021; Eilifsen et al., 2020) by highlighting that institutional logics not only provide prescriptions for DA use but, as perceptual and conceptual filters, may also shape understandings of DA and what it is good for. Below, (the significance of) this is discussed in more detail.

As described in 4.3, in DealCo’s FDD practice, accounting and commercial logic seemed to draw attention to some DA-related action possibilities (e.g. efficient processing) and divert it from other action possibilities (complex modeling) and, thereby, shape the advisors’ conceptual understandings of DA. These understandings, in turn, centered on the efficiency aspect of DA and manifested in an efficiency-centric value creation logic (see Figure 3) (“how I see and define DA is basically big data and how to handle large data sets in an efficient way”). This logic, in turn, seemed to function as a high-level (mental) model of how to use DA in an appropriate and advantageous way – in the context of FDD – and, as such, governed DA use broadly. Theoretically, this links to the idea of “dominant logics” as perceptual and conceptual filters (see Bettis and Prahalad, 1995; Engelmann et al., 2020; Schraven et al., 2015; Von Krogh et al., 2000) found in the management or managerial cognition literature. In that regard, the link between perception and conception warrants some explanation. Dominant or institutional logics provide concepts (e.g. commercial logic → efficiency). These concepts, on the one hand, filter information (i.e. what is perceived) and, on the other hand, help make sense of this information (i.e. how it is conceived) (see Engelmann et al., 2020; Schraven et al., 2015). This may lead to new understandings and concepts – as evident in the case of DealCo’s FDD practice, where it led to an understanding of DA that differed markedly from the “novel insights, better decisions” conceptualizations found in the literature. With respect to institutional research on DA or, more broadly, technology adoption and use in accounting (see Schiavi et al., 2024), this foregrounds the need to pay greater attention to how institutional logics may be implicated in the emergence of conceptual understandings – especially in relation to phenomena that are technology as much as they are ideas. This can contribute to a more comprehensive understanding of the influences of different institutional logics on technological innovation in accounting.

What is important to realize, the significance of the above extends far beyond the theorizing of DA use. It could have direct practical repercussions. If one subscribes to the “novel insight, better decisions” conception of DA, using DA tools to standardize, automate and save time on data preparation tasks likely does not classify as “doing DA.” In contrast, if one subscribes to DealCo’s DA logic (depth, efficiency and effectiveness, see Figure 3), then, using DA to do these very things classifies as “doing DA.” In other words, one’s conception of DA determines if one is – or, more precisely, believes to be – doing DA. This belief, in turn, could have material consequences. In FDD practice, data and analytics were more or less synonymous with digital innovation. Assuming there is a desire to innovate, the conception of DA could have a profound impact on the scope of innovation, e.g. efficiency-oriented uses of DA tools in the context of data preparation versus insight-oriented uses of DA tools and techniques in the context of data analysis. In both cases, DA use would be a testament to digital innovation (for the users). Yet, the scope of innovation would be very different. This points toward alternative conceptions of DA as a powerful but, so far, overlooked hindrances to the adoption and use of more advanced forms of DA. By the same token, a shift in conception – from DA as efficient data processing to DA as insights from big data – could be a powerful driver of further DA-related innovation in accounting practice. To summarize, the above enables a better understanding of the cognitive dimension of DA (non)adoption and (non)use in accounting by drawing attention to some of the more subtle ways in which institutional logics, especially accounting and commercial logic, may be implicated in DA-related action and inaction.

6. Conclusions

This study makes three important contributions to the literature on digitalization in accounting (e.g. Bhimani and Willcocks, 2014; Knudsen, 2020; Möller et al., 2020; Schneider et al., 2015; Spraakman et al., 2021). First, by explicating the links between accounting logic (“technical-professional logic”), commercial logic (“commercial-professional logic”) and the (non)use of DA tools in FDD, it heeds the call for theory- and field-based evidence on (non)adoption and (non)use of DA in accounting for decision-making (see Casas-Arce et al., 2022; Schmidt et al., 2020). For one thing, it suggests that accounting logic and commercial logic can both be drivers of and hindrances to DA use in accounting practice. For another thing, it shows that accounting and commercial logic can be powerful perceptual and conceptual filters that shape accounting professionals’ understanding of DA and, in effect, their use of DA tools. Second, by detailing how accounting logic conflicted, and did not conflict, with DA use in FDD practice, the study adds nuance to the discussions around tensions between the traditional “accounting mindset” and the new DA approach (e.g. Agostino and Sidorova, 2017; Arnaboldi et al., 2017). Third, by highlighting various links between commercial logic and DA use in FDD practice, and by identifying conflicting prescriptions for DA use within commercial logic, the paper extends prior discussions about the business value of DA in accounting (e.g. Brands and Holtzblatt, 2015; Nielsen, 2018).

This study has important implications for FDD and, more broadly, accounting practitioners. First, in light of the vehement calls for swift and thorough DA adoption (“embrace or vanish”), our study suggests that a more measured approach might be the better option in accounting for decision-making. While efficiency-related DA opportunities should be exploited, more caution is needed in relation to insight- and foresight-related DA opportunities. In high-stakes decision-making, more complex forms of DA, e.g. predictive analytics, may not give that much comfort to executive decision-makers and, in effect, only create little value for them in these situations. Second, maintaining a traditional “accounting mindset” might, to some extent, be conducive to DA-based value creation. Indeed, thanks to its discriminatory function (see reconciliations and data integrity), accounting logic can be a useful guidepost for determining which of the many DA opportunities to pursue and which to forego. Third, given the link between conceptions of DA and the use of DA, altering these conceptions among DA users, i.e. through reeducation, could be a powerful lever for driving DA-related change – in case that was desired.

This study has some limitations. First, deal-making and data-sharing practices vary across markets. In the USA, for instance, where there is little sell-side preparation, sharing transactional data is more common than in Northern Europe. Because of these differences, some of our findings may not replicate in other geographical settings. Furthermore, we only studied one organization which makes it difficult to assess the generalizability of our findings. In that regard, one should also consider that in M&A decision-making, stakes, visibility and demand for justification are high. Thus, our findings likely transfer well to decision-making contexts in which the stakes and demand for justification are equally high (e.g. other strategic investment decisions) but might not transfer as well to decision contexts that lack these characteristics (e.g. more operationally-oriented decisions). Moreover, this is a study of early-stage DA adoption. At a later adoption stage, some of the findings could have looked different.

Finally, we propose two opportunities for further research. On the one hand, future studies could choose a comparative approach and investigate how the links between accounting and commercial logic and DA use may vary across different accounting firms and regions. On the other hand, future research could leverage longitudinal designs and investigate which of the identified links persist (break down) as the result of prolonged and intensified DA use. Either way, considering FDD’s unique place at the intersection of financial accounting, management accounting and auditing, we hope that our study will motivate other researchers to harness the potential of FDD as a context to learn about digitalization in accounting.

Figures

Figure 1.

Accounting logic as hindrance and driver of DA use

Figure 2.

Commercial logic as driver and hindrance of DA use

Figure 3.

DealCo’s DA logic

Figure 4.

DA use in DealCo’s FDD practice

Table 1.

Overview of primary data

Data source	Position			ID	Minutes
Interview	Partner, Director			P1	60
				P2	60
				P3	50
	(Senior) Manager			M1	60
				M2	60
				M3	60
				M4	60
	(Senior) Analyst/Associate, other staff			A1	60
				A2	40
				A3	50
				A4	60
Follow-up interview				M1	30
				M4	30 + 30
Walkthrough session				M4	60
				M4	30
Kick-off meeting				P1, M4	30
Close-out meeting				P1, M1, M4	80

Notes:

To preserve the anonymity of the participants, we use three broader categories: 1) Partner and Director, 2) (Senior) Manager and 3) (Senior) Analyst/Associate and other staff. In some firms, the junior staff is called “Analysts,” in others “Associates”

Source: Table courtesy of Kastrup et al. (2024)

Table 2.

Overview of secondary data

Source	Quantity
White papers	12
Company videos	10
Website posts	6
Presentations	2
Podcasts	2
Webcasts	1

Source: Table created by the authors

Note

1. Kastrup et al. (2024) use the same empirical case to investigate how practical and theoretical judgments are invoked in data-driven FDD (drawing on Dewey’s theory of inquiry) and offer complementary insights into the role, use, and importance of accounting professionals’ judgment in data-driven inquiries.