Pathways to Value from Business Analytics

Through what pathways does business analytics (BA) contribute to business value? To answer this question we argued that BA tools and capabilities only produce value if they are used, so we set out to explore different types of BA use. This led to the identification of two types of BA users—analytics professionals and analytics end-users (from executives to the shop floor)—which in turn led to identification of the three “pathways to value from business analytics”, namely provision of advisory services, creation and enhancement of BA tools and the BI-platform, and use of BA tools by end users. As a preliminary empirical assessment of the validity of these three pathways, we conducted eleven one-hour interviews with thirteen senior managers with a wide range of interests in BA. Results from those interviews are consistent with our claim that the three pathways exist and are important sources of business value from BA. Keywords: Business analytics (BA), business intelligence (BI), business value, business benefits, success factors, advisory services, decision-making, BA tools Knowledge Management and Business Intelligence 2 Thirty Fourth International Conference on Information Systems, Milan 2013 Introduction Although there are many definitions of both “business analytics” and “business intelligence”, in this paper we define business analytics (BA) as the use of data to make sounder, more evidence-based business decisions, and business intelligence (BI) as IT-based BA tools, e.g., statistical and quantitative techniques, explanatory and predictive models, data warehouses, on-line analytical processing (OLAP), visualization, and data mining that enable BA (Howson 2011; Negash 2004). In the past decade, there has been massive interest worldwide in BA and therefore BI. As evidence, BI topped the list of “Technical priorities for CIOs” in Gartner’s annual global surveys of CIOs in the three of the five years, 2007-2011 (Hagerty et al. 2012, p. 47). Further, the spate of multi-billion dollar takeovers of BI firms in the past five years, e.g., by Oracle (of Hyperion), IBM (of Cognos and SPSS), and SAP (of BusinessObjects), as well as various vendors’ current touting of their in-memory database technologies (IBM 2013; SAP 2011) suggests that vendors believe that BA is likely to make a major contribution to firm performance in the coming decade. As a way of framing this study, we have reproduced Seddon et al.’s (2012) recent process model of business-analytics use in organizations in Figure 1. According to those authors, the left-hand side of their model “relates to the use of business-analytic capabilities to produce information and insight”, and the right-hand side “relates to the use of the organization’s entire set of capabilities to produce business value” (p. 3). In this paper, we focus on both (a) the Use Analytic Capabilities concept on the left of Figure 1, and (b) two distinct types of business analytics users, namely (i) Analytics Professionals and (ii) Analytics End-Users (who include both the Analytical Executives and Analytical Employees in Figure 1). However, although we focus on BA users, we also recognize that BA use, per se, is not a source of organizational benefits. Rather, it is only when insights from BA use result in decisions and actions, such as those depicted on the right of Figure 1, that benefits from BA start to flow. Thus, in the context framed by Figure 1, the research question we seek to answer in this paper is: Through what pathways does business analytics contribute to business value? Process Model (executed over and over again in different parts of the organization) Organizational Benefits from Analytics Use, from the perspective of Senior Management enable Insight(s) Decision(s) Competitive Actions that change the organization’s capabilities Use Analytic Capabilities Analytical Capabilities Analytical People • Analytical executives • Analytical professionals • Analytical employees Enabling Technology • High-­‐quality data • Integrated Business-­‐ Intelligence Platform Competitive Actions that use the organization’s existing capabilities Path 1 Organizational Capabilities enable = results in Path 3: results in changes in, e.g., learning Path 2: results in changes in Path 2 Path 1 Figure 1. A Business-Analytics Process Model (Source: Seddon et al. 2012) Perhaps one obvious answer to this question is to point to the three pathways depicted in Figure 1. However, these pathways—with their focus on different types of competitive actions—are not the topic of this paper. Rather, our interest in this paper is in different types of BA users mentioned above and the different roles they play in BA use. Figure 1 actually says very little about BA use. For example, it says that BA use is enabled by Analytical Capabilities, and results in Insights that lead to Decisions, and so on, but Tamm et al. / Pathways to Value from Business Analytics Thirty Fourth International Conference on Information Systems, Milan 2013 3 it is silent about the nature of BA use and users. Yet, BA use is obviously key to realizing benefits from BA: because no use means no benefits! Further, it is clear there are many different types of BA use (e.g., drilling down for information on an active dashboard application on an iPad is very different to heavyduty statistical analysis using a package such as R). Logic suggests that these different types of BA use and users probably contribute to organizational benefits in very different ways. Therefore, our goal in this paper is to present a much richer description of BA users and BA use than that depicted in Figure 1, and to apply insights from this richer description to identify the primary pathways through which different types of BA use contribute to business value. To achieve these goals, this paper is structured in two parts. First, we present grounds for believing that there are three main pathways through which BA users contribute to realization of business value from BA. Second, to assess the validity of these three pathways, we conducted eleven interviews with thirteen business-analytics experts in Australia. Interviewees were thought leaders with a wide range of involvement with analytics. Our preliminary empirical findings, reported in the second half of this paper, suggest that each of the three pathways is an important contributor to business value from BA. Two Types of Business Analytics Users Surprisingly little has been written about the different types of BA users, and the different roles they play in BA use. Apart from early articles distinguishing between hands-on and chauffeured use of Decision Support Systems, e.g., Keen (1980), the only articles that we could find that addressed the issue directly were Davenport and Harris (2007), Davenport et al. (2010), Davenport and Patil (2012), and Accenture (2013). Based on their views, as well as the existence of professional analytics organizations such as The Data Warehousing Institute, the Institute of Analytics Professionals of Australia (IAPA), R user groups, and so on, we argue that there are two distinct types of BA users: Analytics Professionals (APs) and Analytics End-Users (AEUs). These two types of BA users, and the characteristics of their BA use, are described in Table 1 and discussed in more depth in the sections that follow. Table 1. Two Main Types of Analytics Users BA User Description Characteristics of BA Use Analytics Professional (AP) Davenport and Patil (2012) call analytics professionals “data scientists”. Typically, they provide evidence-based insights on a range of structured and unstructured questions to an organization’s more senior managers. They also guide the embedding of such insights into operational systems. APs use analytic tools and the scientific method to build (i.e., discover) more accurate understandings of cause and effect for organizationally relevant phenomena, such as revenue and cost streams. This may involve the use of all types of quantitative and qualitative techniques, but in particular, it includes the use of powerful tools for interactive visualization, predictive modeling and data mining, and prescriptive modeling, simulation, and optimization techniques. Analytics End-User (AEU) Analytics end-users are business users throughout the organization, from senior executives down to the shop floor. Such people typically have good business knowledge, but frequently do not have strong statistical or analytic skills. AEUs use capabilities built into BI software to inform organizational decision-making. This may involve the use of conventional reports, spreadsheets, dashboards, scorecards, online-analytical processing (OLAP), and ad hoc queries delivered either to the desktop or via mobile devices. Insights from APs, e.g., decisions about market segmentation, may frame the way that information is presented to AEUs. Knowledge Management and Business Intelligence 4 Thirty Fourth International Conference on Information Systems, Milan 2013 It is important to point out that BA usage can be classified in various ways, and that our distinction in Table 1—which involves classifying usage by user type—is only one of those ways. For example, the nature of the underlying problem that a BA user is trying to solve can be used as a basis for classifying BA use, e.g., whether the problem is structured or unstructured; or operational or strategic in nature (Gorry and Scott Morton 1971). Although it is difficult to generalize, we would expect APs to be more frequently engaged in solving unstructured and strategic problems, and AEUs with structured problems (whether AEUs’ analytic focus is strategic or operational depends on the position of the end-user in question). Another example is the historical-to-predictive BA classification proposed by Gartner (see Table 2). According to Gartner, in 2012 “most users still focus on measurement of the past, with only 13 percent of users making extensive use of predictive analytics. Less than 3 percent use prescriptive capabilities such as decision/mathematical modeling, simulation and optimization” (Robb 2012). In terms of Gartner’s classification, we would expect that APs would be more likely involved with the more-advanced Diagnostic, Predictive, and Prescriptive analytics in Table 2, and AEUs would be more likely involved with the lessadvanced Descriptive and Diagnostic Analytics types of BA. All of the approaches described above have their own merits and provide useful, complementary lenses to examining BA and BA benefit realization. In this paper, we have chosen to focus on the BA-user-based categorization, as we believe this largely unexplored approach holds strong promise in improving and enriching the understanding of how BA benefits unfold. As benefits from BA must flow from BA use, it is important to understand how people in organizations use BA and to identify the various roles associated with BA use that, ultimately, contribute to business value. Table 2. Gartner’s Four Types of Analytics Use Category Definition (Source: Schlegel et al. 2013, p. 3) Descriptive “The vast majority of applications built with BI and analytics platforms to date could be labeled "descriptive" because critical capabilities, such as reports and dashboards, are used to describe the dimensions and measures of a particular aspect of the business. So, for example, a measure such as on-time delivery could be defined in a well-governed data model and enable users to report on the goal and actual value for that measure by various dimensions, such as customer segments or time periods.” Diagnostic “Increasingly, Gartner sees more organizations building diagnostic analytics that leverage critical capabilities, such as interactive visualization, to enable users to drill more easily into the data to discover new insights. For example, visual patterns uncovered in the data might expose an inconsistent supply chain process that is the root cause of an organization's ability to consistently reach its goal for on-time delivery.” Predictive “As organizations mature at diagnostic analysis, they become so adept at understanding the root causes in their business processes that they can identify the explanatory variables that predict what the measure will be in a future period. For example, a predictive analytic system could be built to forecast the on-time delivery measure.” Prescriptive “Solutions can be further evolved to prescriptive analytics as the insights from predictive models are integrated into business processes to take corrective or optimal actions.” It is also important to point out that all BA use relies on the ready availability of both (a) BI platforms and tools within the organization (e.g., data warehouses, data marts, extract-transform-load (ETL) processes, toolsets such as those from both major vendors and open-source providers, and mobile-informationdelivery solutions), and (b) relevant high-quality data. Without ready access to both types of resource, BA efforts will be hamstrung and ineffective. Our objective in this paper is to show that the two types of BA use highlighted in Table 1 require very different human capabilities, and produce benefits through very different mechanisms. With that goal in mind, the two types of BA users and use summarized in Table 1 are now discussed in turn. Tamm et al. / Pathways to Value from Business Analytics Thirty Fourth International Conference on Information Systems, Milan 2013 5 Analytics Professionals (APs) As explained in Table 1, APs major focus is on the use of analytic tools to build more accurate understandings of cause and effect in the organization and its environment. Stated differently, the goal of APs is to use scientific method (Popper 1959) and abduction (Peirce 1903) to build causal models of phenomena that are of interest to the organization, e.g., the drivers of organizational revenue and cost streams. Therefore, APs are sometimes referred to as data scientists (Davenport and Patil 2012). In terms of competencies, APs are very similar to Ph.D. researchers as they need to have (a) a deep understanding of the context of enquiry, (b) a deep understanding of goals and limitations of cause-andeffect explanations of empirical phenomena (e.g., does smoking cause cancer?), (c) strong computing and data-manipulation capabilities, (d) competence with statistical analysis, and (e) the ability to communicate with business managers to explain how insights from the data potentially impact the business. Although AP duties have been performed in business for many years, e.g., management accountants performing financial modeling to support, say, merger-and-acquisition decisions, the AP as a specialized position is relatively new to business. As a result, there are few degree programs that seek to produce APs.1 The roles APs play in BA use may be divided into at least the four shown in Table 3. The first two roles in Table 3 may be termed Advisory, because they involve providing advice to senior business managers who make the actual decisions. (APs do not make these decisions.) The third and fourth role are concerned with building BA capabilities for the use of both APs and AEUs. Here, the AP has much more say in decision-making (although large investments, say, in a SAP HANA or IBM Netezza platform, would of course be subject to normal IT-governance procedures (Weill and Ross 2004)). Of the two latter roles, the “Supervising Development” role involves working with IT-development project teams to develop workable tools or products, e.g., dashboards, for both routine and ad hoc use by business users. In some cases, this will involve embedding insights from advisory work, such as customer segmentation insights, into operational processes, such as distinct types of marketing campaigns. Table 3. Professional Analyst (AP) Roles