Printer Friendly

Exploring a measurement of analytics capabilities.

Changes in the volume and velocity of data have led many organizations to consider assessing and improving analytics capabilities. The purpose of this research is to describe a methodology developed to assess organizations' analytics capabilities and explore the empirical value of data collected using this methodology. The measurement for analytics capabilities was developed by IBM during 2009-11 marketing efforts. To assess the data's empirical value, we investigate whether measurements of analytics capabilities are internally consistent, associated with decisions to invest in analytics software and hardware, and able to explain firm profitability. In analyzing consistency, we find a natural sequence in the development of analytics capabilities. Exploring decisions to invest in analytics, we discover that firms with higher levels of capabilities are more likely to invest, as are firms that are larger and located in more profitable industries. However, we find no relationship between analytics capabilities and firm profitability.

Business Economics (2016) 51, 27-35. doi: 10.1057/be.2016.9

Keywords: analytics capability, IT management, analytics investment decisions, firm profitability


The rapid changes in the volume, variety, and velocity of available data have led many organizations to consider first assessing and then improving their analytics capabilities. Analytics can be defined as "the extensive use of data, statistical and quantitative analysis, explanatory and predictive models, and fact-based management to drive decisions and actions" [Davenport and Harris 2007, p. 7], Analytics capabilities (a subset of IT capabilities) relate to the ability to capture, store, manage, and analyze data. (1) The emergence of "big data"[Rogers 2011; Davenport, Barth, and Bean 2012] as an important issue in managing organizations has further stimulated interest in improving analytics capabilities. As a result, increasing numbers of organizations view analytics capabilities as strategic and a source of competitive advantage [Davenport and Harris 2007].

At the micro level, those who develop and market the tools of the trade (such as IBM, SAP, SAS, and Oracle) need to understand their clients' current capabilities in order to identify gaps that could be productively filled. In a larger framework, any scientific attempt to explore and quantify the link between a firm's performance and its analytical capabilities requires an appropriate yardstick for the measurement of those capabilities.

A recent preliminary effort in a related area suggested an approach to assessing maturity of business intelligence, without exploring the link between maturity and firm performance [Tan, Sim, and Yeoh 2011]. More complete efforts have been made in this direction within the larger framework of information technology (IT) [Aral and Weill 2007; Stoel and Muhanna 2009]. The latter emphasize the importance of moving beyond aggregate overall measures of IT capability and focusing on IT capability by specific type and nature.

It is in this spirit that our paper explores the nature and usefulness of one such measurement tool for analytics capabilities, developed and used internally by IBM during its marketing efforts between 2009 and 2011. (See Appendix Section A.1 for an explanation of the authors' research relationship with IBM.) We investigate whether these measurements of analytics capabilities are internally consistent, associated with decisions to invest in analytics software and hardware, and able to help explain firm profitability.

1. Background

Between 2009 and 2011, as part of its Smarter Analytics efforts, IBM offered workshops for organizations to analyze, strategize, and plan for investments in analytics. These organizations included government agencies, nonprofits, and privately held and publicly traded for-profit organizations. During each engagement, IBM gathered details on clients' assessments of current analytics capabilities and future goals for the purpose of providing recommendations for clients' information agenda.

In the methodology used by IBM, analytics capabilities fell into four categories: information foundation (IF), trusted information (TI), define and govern (DG), and analytics and optimization (AO). These four capabilities represent categorization by functionality, similar to the IT capabilities found in Tan, Sim, and Yeoh [2011] and several of the categories in Weill, Subramani, and Broadbent [2002], rather than organized around external vs. internal focus [Day 1994; Stoel and Muhanna 2009].

For each of these categories, IBM developed a series of statements pertaining to functionalities within the category. (See Appendix Section A.2 for a listing of these statements.) Representatives of each client were asked to evaluate each statement on a 0-10 scale, where 0 meant "rarely or never true" (weakest level of capability) and 10 meant "consistently true" (strongest level of capability). Average scores for each of the four main capability categories were stored in a database that IBM provided to the authors, along with other information about the clients.

Given the subjective nature of the assessment process, one might wonder about the usefulness of the resulting capability measures. We explore the empirical value of the methodology and the data by checking for (1) internal consistency among measurements, (2) an ability to predict clients' subsequent decisions to purchase IBM products, and (3) an effect on clients' subsequent profitability. Although the data perform reasonably in the first two contexts, we have been unable to establish a significant link between the subjective assessment of capability and subsequent profitability.

2. Internal Consistency of Capability Measurements

Inspection and logic suggest a natural progression for the development of analytic capabilities defined by IBM. Being able to capture, store and manage data (IF) would by definition be truly foundational. A strong foundation would logically be a precursor to having "mature" architecture and "mature" processes (DG) as well as being able to work "easily, reliably, seamlessly, automatically and intelligently" with information (TI). Capability in analytics and optimization (AO) would be hampered if these other capabilities were weak; as Davenport, Harris, and Morison [2010, p. 19] indicate: "Good data is a prerequisite for everything analytical."

To check for internal consistency among the four IBM capability measurements, we test the following hypotheses:

H1: A higher level of IF capability supports a higher level of TI capability (all else held constant).

H2: A higher level of IF capability supports a higher level of DG capability (all else held constant).

H3: Higher levels of IF capability, TI capability, and DG capability support higher levels of AO capability (all else held constant).

We control for differences in industry, geography, and year in this investigation, since variation in these factors could contribute to variation in capability measurements.

Each of the three hypotheses was tested using the results of a seemingly unrelated regression estimation (SURE) technique, applied simultaneously to a set of three equations:

DG = [f.sub.DG](industry, geography, year, IF) + [[epsilon].sub.DG], (1)

TI = [f.sub.TI](industry, geography, year, IF) + [[epsilon].sub.TI], (2)

AO = [f.sub.AO]{industry, geography, year, IF, DG, TI) + [[epsilon].sub.AO]. (3)

A regression technique allows us to estimate the individual contribution of each determinant of a capability measure, all else held constant. Given the recursive nature of the equations, SURE is more appropriate than ordinary least squares (OLS) whenever the stochastic error terms ([epsilon]) in the equations are correlated [Gujarati 1995, p. 682], A Breusch-Pagan test for independent error terms across the three equations indicated that the error terms of these three equations were in fact correlated, with a p-value of 0.000 [Breusch and Pagan 1980; StataCorp 2009, p. 1841],

Table 1 presents abbreviated estimation results for 382 organizations. It shows that Hypotheses 1 and 2 are fully supported. Organizations have significantly higher current capability in TI and DG, the higher their capability in IF: all other things being the same, the assessments of capability in TI and DG are roughly 0.7 points higher, on average, for every point higher the assessment of capability in IF. Hypothesis 3 appears to be partially supported. Variation in current capability in AO, the ultimate goal of analytics maturity, is significantly linked to variation in current capabilities in the intermediate stages (TI and DG) of development, but not significantly linked to variation in current capability in the initial stage (IF) of development. However, combining the results across the three hypotheses, one can see that the effect of IF on AO is indirect, occurring via its effect on TI and DG. All other things being equal, for every additional point of capability in TI (DG), the assessment of capability in AO is on average about 0.55 (0.16) points higher. The differential in impact between capabilities in TI and DG is statistically significant (p-value = 0.000).

The progression or sequencing of analytics capabilities hypothesized in this research is similar in concept to the "powerful chain of influence among the capability types" found for IT [Kim and others 2011, p. 501). Although Fink [2011] defined a set of IT capabilities that differ from those of Kim and others [2011], he also found a sequencing effect. Our findings concerning the progression of analytics capabilities are consistent with the idea that developing information infrastructure first and then addressing data governance and quality issues are important precursors for obtaining an analytics capability.

Thus, our results are generally consistent with expectations and appear to raise no "red flags" concerning the nature of the IBM capability measurements. It must be noted that, although the IBM data covered a wide range of firms and organizations of various sizes and geographical locations, the estimation results might have been somewhat different if the data had been a completely random sample of firms and organizations in the marketplace.

3. Ability to Predict Decision to Make Purchase

A more important question about the capability measurements developed by IBM is whether they are robust enough to serve a practical purpose. To address this question, we investigate whether the measurements were useful in predicting clients' decisions to purchase products. Additional information from IBM allowed us to determine which clients purchased IBM products after their workshops with IBM (through mid-2012).

For guidance in modeling the decision, we reviewed related literature on investment decisions. A literature search revealed only one exploratory study that investigated the factors that drive investment in analytics [Xavier, Srinivasan, and Thamizhvanan 2011], However, since analytics can be thought of as a component of IT, we expanded our literature review to include research that addressed investment in IT in general. It seems reasonable to believe that factors that influence investment in IT in general might also influence investment in analytics in particular.

In a survey of 84 Indian enterprises, Xavier, Srinivasan, and Thamizhvanan [2011] found that investment in analytics is positively affected by prior investment in analytics. This evidence provides the basis for our key hypothesis that the capability measurements developed by IBM would be positively related to the likelihood that clients purchase IBM analytics software and hardware. In our analysis, we used a measure of total capability (IF+DG+TI+AO) to reflect prior investment in analytics.

Guided by the literature, we selected three other factors that would be likely to influence investment in analytics: firm size, organizational slack, and industry competitiveness.

Firm size is repeatedly mentioned as a factor influencing IT investment. Mitra and Chaya [1996] suggested that larger firms might invest in IT to achieve economies of scale and improve coordination. Indeed, Ravichandran, Han, and Hasan [2009] found firm size to be positively related to IT investment, as did Giunta and Trivieri [2007], Xavier, Srinivasan, and Thamizhvanan [2011] reported a similar result for investment in analytics. In our analysis, we used the natural logarithm of the firm's total revenues to represent firm size.

A strong case for including organizational slack as a factor influencing IT investment is made by Ravichandran, Han, and Hasan [2009]. Organizational slack is defined as the availability of resources above and beyond those necessary for meeting immediate business requirements [Cyert and March 1963; Nohria and Gulati 1996], Ravichandran, Han, and Hasan [2009] argued that firms that have more slack have better access to the funds needed for IT investment and indeed found that highly leveraged firms (having lower levels of organizational slack) tended to have lower levels of IT investment. In our analysis, we used the firm's debt/equity ratio to represent organizational slack (noting that increases in the debt/equity ratio are associated with decreased organizational slack).

Controlling for industry appears to be commonplace in studies of IT investment. Ravichandran, Han, and Hasan [2009] included sector dummy variables, while other researchers restricted themselves to studying a single sector or type of sector [Steiner and Teixeira 1990; Harris and Katz 1991], To the extent that analytics can be thought of as an innovative form of IT, the work of Aghion and others [2005] suggests that industry competitiveness in particular might influence the investment decision. These authors presented theoretical arguments and empirical evidence that innovation first increases but eventually decreases as an industry moves toward perfect competition. In light of this aspect of the literature, we opted to include a measure of industry competitiveness in our model of the investment decision. We followed the example of Aghion and others [2005], who used a measure of industry competitiveness based on the industry mean value of operating profit minus financial cost (as a percent of sales). Given that their findings were robust to excluding financial cost, we chose to measure industry competitiveness using the industry median net income margin, as a percent of net sales (noting that increases in the industry median net income margin are associated with decreases in industry competitiveness).

To the extent that these measurements for firm size, organizational slack, and industry competitiveness were available, we supplemented the records in the IBM database with corporate information from public sources (specifically, S&P Capital IQ and Bloomberg). Unfortunately, these data were largely unavailable for privately held corporations, nonprofit organizations, and government agencies. Though the number of usable observations shrank accordingly in this analysis, the sample size remained large enough to support our statistical analysis.

We used this specification to estimate a binary logit model with decision to invest as the dependent variable, in order to determine whether the IBM measurements of analytics capabilities could be useful for predicting these decisions. The model was estimated with and without dummy variables reflecting the firm's geographical location as additional controls. The results of this estimation are presented in Table 2.

Most notably, the significant coefficient associated with the client's self-assessed level of analytics capability suggests that the IBM measurement of capability in fact did have a meaningful weight in predicting the client's decision to purchase analytics software and hardware from IBM within the given time frame. In line with the results of Xavier, Srinivasan, and Thamizhvanan [2011], the estimation results suggest that a firm with a greater level of analytics capability is more likely to invest in additional analytics.

Of the other three independent variables included in the model, only firm size and industry competitiveness demonstrated significant effects. As expected, estimation results suggested that larger firms were more likely to invest in analytics. The results also suggested that firms in less competitive industries were more likely to invest in analytics, reflecting one portion of the inverted-U relationship discussed by Aghion and others [2005]. (2) These results could be associated with expenditures on analytics being a relatively larger burden in terms of overhead for firms that are small in terms of physical size or profit margin.

An alternative explanation of our results concerning the influence of industry competition follows from Ravichandran, Han, and Hasan [2009], who found that firms tended to imitate the IT investment intensity of their peers. Perhaps firms in more profitable industries will seek to imitate what they believe or know to be the high propensity for analytics investment in their industry. This behavior is consistent with choosing to invest in an industry that is more profitable, assuming that firms view investment in analytics as a pathway to greater profits.

Insignificant results for organizational slack as represented by the firm's debt/equity ratio were somewhat surprising to us, given the evidence presented by Ravichandran, Han, and Hasan [2009], However, the insignificance of this variable would probably come as no surprise to Bourgeois [1981, p. 30], who expressed the opinion that, though "certain public financial data" might serve "as a useful first cut in measuring slack," better measures could be had if researchers were able to "penetrate the organization's boundary to collect data from its managers."

The estimated logit results were able to predict correctly the investment decisions in 72.1 percent of firms in the sample. Only a small portion (2.7 percentage points) of the percentage correctly predicted came from the inclusion of the geographical controls. Omitting the organizational slack variable yielded very similar estimation results. (3)

Given the nature of the data used in this portion of our analysis, it must be noted that the estimation results may not apply to nonpublicly traded for-profit corporations.

4. Role in Explaining Profitability

A more advanced test of the usefulness of the IBM measurements of analytics capabilities lies in a search to find a link between firms' subjective assessments of analytics capabilities and subsequent profitability, a measurement of firm performance. The possibility of finding a link follows from the direct and indirect linkages found between IT capabilities and firm performance in some studies [Kim and others 2011]. To extend our search for the value of IBM capability measurements, we took a modest stab in this direction.

We were somewhat encouraged in this effort by the work of Brynjolfsson, Hitt, and Kim [2011], These authors created an index for data-driven decision making, based on senior executives' responses to three survey questions concerning the extent to which data were available and used in firms' business decisions. These authors used regression analysis to find a significant positive link between their index (an independent variable) and two of three measurements of business profitability (the dependent variable), holding constant certain other firm characteristics gleaned from the survey questionnaire. Additional regression analyses to explore the direction of causality were less definitive, leaving the authors unable to make inferences on the matter.

Our efforts used a firm's net income margin (as a percent of net sales) as the measurement of a firm's profitability. In our attempt to explain profitability, we specified a sparse model, given our data limitations and our narrow interest in the role of analytics capabilities. Specifically, we chose to focus on whether a measure of analytics capabilities could add explanatory power to a regression model relating a firm's profitability to the median profitability in its industry. Analytics capabilities were measured in the year of the IBM workshop; profitability was measured one year later. We initially used a measure of a firm's total capability (IF+DG+TI+AO) to reflect analytics capabilities in this phase of our analysis. To acknowledge the industry-specific setting of the profitability data, we subsequently used a dummy variable indicating whether the firm had analytics capabilities higher than average among its industry peers within our sample.

The results of these efforts are shown in Table 3 for a sample of publicly traded for-profit firms. As expected, all three estimations confirmed that a firm's profitability is significantly and positively related to the median profitability within its industry. However, neither indicator of analytics capability performed well when added as an explanatory factor. It was not apparent from our work that the IBM measurement of analytics capabilities had value as a predictor of a firm's profitability. The general economic turmoil in the economy during the time period being studied could have obscured the relationship.

5. Summary and Conclusions

In this paper, we have (1) described a methodology used by IBM to measure the analytics capabilities of its clients between 2009 and 2011 and (2) demonstrated some empirical value of data that measure analytics capabilities.

The IBM approach yielded measurements that we found to be logically related to one another. Specifically, organizations have significantly higher current capability in intermediate analytics competencies (TI and DG), the higher their capability in basic analytics competencies (IF). Variation in current capability in advanced analytics competencies (AO) is more significantly linked to variation in current capabilities in TI and DG, rather than in IF. Combining the results suggests that the effect of IF on AO is indirect.

The measurements were also found to have had some internal value in predicting whether clients responded to IBM marketing efforts by purchasing IBM products in the near future, thus making the measurements potentially useful in guiding future marketing efforts. Specifically, the results showed that firms that have higher current levels of analytics capabilities are more likely to invest in analytics. Firms that are larger and located in less competitive (more profitable) industries are also more likely to invest in analytics. These results suggest that firms tend to build on their analytics capabilities, since they have likely already experienced the value of their past investments. Interestingly, no relationship between analytics investment and organizational slack was found at the firm level.

A link between a firm's analytics capabilities and its subsequent profitability was not found, as was the case in some of the related studies on IT capabilities (Kim and others 2011]. Extensions of this analysis to investigate profitability over a longer term may well be particularly appropriate, given the atypical state of the economy between 2010 and 2012.


The authors would like to acknowledge and thank IBM for supporting this research, and would especially like to recognize Howard Fields, Carolyn Martin, Will Reilly, Michele Shaw, and Rebecca Shockley of IBM for their assistance and contributions.


Aghion, Philippe, Nick Bloom, Richard Blundell, Rachel Griffith and Peter Howitt. 2005. "Competition and Innovation: An Inverted-U Relationship." The Quarterly Journal of Economics, 120(2): 701-28.

Aral, Sinan and Peter Weill. 2007. "IT Assets, Organizational Capabilities, and Firm Performance: How Resource Allocations and Organizational Differences Explain Performance Variation." Organization Science, 18(5): 763-80.

Bourgeois, L. Jay 1981. "On the Measurement of Organizational Slack." The Academy of Management Review, 6(1): 29-39.

Breusch, Trevor and Adrian Pagan. 1980. "The Lagrange Multiplier Test and its Applications to Model Specification in Econometrics." Review of Economic Studies, 47(1): 239-53.

Brynjolfsson, Erik, Lorin Hitt and Heekyung Hellen Kim. 2011. Strength in Numbers: How Does Data-Driven Decisionmaking Affect Firm Performance? Social Science Research Network (April 22). 1819486 (accessed April 21, 2013).

Cyert, Richard and James March. 1963. A Behavioral Theory of the Firm. Prentice-Hall.

Davenport, Thomas and Jeanne Harris. 2007. Competing on Analytics: The New Science of Winning. Harvard Business School Press.

Davenport, Thomas, Jeanne Harris and Robert Morison. 2010. Analytics at Work: Smarter Decisions, Better Results. Harvard Business School Press.

Davenport, Thomas, Paul Barth and Randy Bean. 2012. "How "Big Data" is Different." Sloan Management Review, 54(1): 43-6.

Day, George. 1994. "The Capabilities of Market-Driven Organizations." Journal of Marketing, 58(4): 37-52.

Fink, Lior. 2011. "How Do IT Capabilities Create Strategic Value? Toward Greater Integration of Insights from Reductionistic and Holistic Approaches." European Journal of Information Systems, 20(1): 16-33.

Giunta, Anna and Francesco Trivieri. 2007. "Understanding the Determinants of Information Technology Adoption: Evidence from Italian Manufacturing Firms." Applied Economics, 39(10): 1325-34.

Gujarati, Damodar. 1995. Basic Econometrics, 3rd ed. McGraw-Hill.

Harris, Sidney and Joseph Katz. 1991. "Firm Size and the Information Technology Investment Intensity of Life Insurers." MIS Quarterly, 15(3): 333-52.

Kim, Gimun, Bongsik Shin, Kyung Kyu Kim and Ho Geun Lee. 2011. "IT Capabilities, Process-Oriented Dynamic Capabilities, and Firm Financial Performance." Journal of the Association for Information Systems, 12(7): 487-517.

Mitra, Sabyasachi and Antoine Chaya. 1996. "Analyzing Cost-Effectiveness of Organizations: The Impact of Information Technology Spending." Journal of Management Information Systems, 13(2): 29-57.

Nohria, Nitin and Ranjay Gulati. 1996. "Is Slack Good or Bad for Innovation?" The Academy of Management Journal, 39(5): 1245-64.

Ravichandran, T. Ravi, Shu Han and Iftekhar Hasan. 2009. "Effects of Institutional Pressures on Information Technology Investments: An Empirical Investigation." IEEE Transactions on Engineering Management, 56(4): 677-91.

Rogers, Shawn. 2011. Big Data is Scaling BI and Analytics. Information Management (September 1), http://www. (accessed February 8, 2013).

Sambamurthy, Vallabh and Robert Zmud. 2000. "Research Commentary: The Organizing Logic for an Enterprise's IT Activities in the Digital Era - A Prognosis of Practice and a Call for Research." Information Systems Research, 11(2): 105-14.

StataCorp, 2009. Stata Base Reference Manual: Release II. StataCorp LP.

Steiner, Thomas and Diogo Teixeira. 1990. Technology in Banking: Creating Value and Destroying Profits. Dow Jones-Irwin.

Stoel, M. Dale and Waleed Muhanna. 2009. "IT Capabilities and Firm Performance: A Contingency Analysis of the Role of Industry and IT Capability Type." Information & Management, 46(3): 181-9.

Tan, Chee-Sok, Yee-Wai Sim and William Yeoh. 2011. A Maturity Model of Enterprise Business Intelligence. Communications of the IBIMA. http://www.ibima html (accessed July 24, 2014).

Weill, Peter, Mani Subramani and Marianne Broadbent. 2002. "Building IT Infrastructure for Strategic Agility." MIT Sloan Management Review, 44(1): 57-65.

Xavier, Maria Joseph, Anil Srinivasan and Arun Thamizhvanan. 2011. "Use of Analytics in Indian Enterprises: An Exploratory Study." Journal of Indian Business Research, 3(3): 168-79.


Section A. 1 describes the relationship between the authors and IBM. Section A.2 provides the statements used by IBM in scoring analytics capabilities.

A.1. History of research relationship with IBM

In 2011, an IBM executive was on the Advisory Council of our university's Center for Business Analytics. When he heard that we were interested in research projects involving analytics, he began to explore possibilities for us at IBM. He put us in touch with an IBM group involved in collaborating with academic researchers. This group connected us with IBM's Smarter Analytics marketing team members, who had collected analytics data and were interested in seeing what could be learned from the data. We proposed several project ideas that were of interest to IBM. On January 20, 2012, the Center for Business Analytics signed an agreement on the exchange of confidential information. We shared the results of our analyses with IBM, and we were given permission to publish the results of our work.

A.2. Statements used by IBM in scoring analytics capabilities

Representatives of each client were asked to evaluate each statement on a 0-10 scale, where 0 meant "rarely or never true" (weakest level of capability) and 10 meant "consistently true" (strongest level of capability).

Information Foundation (IF)

Data management

* We use a cost effective, high performance, highly available relational database management system.

* For real-time applications where speed is critical, we use an in-memory database system that performs far faster than conventional disk-based databases.

* We perform high-volume, low-latency replication for business continuity.

* We have an enterprise approach to archive historical transaction records from ERP, CRM, and custom applications and store them securely and cost-effectively.

* We can create a 'production-like' test environment that accurately reflects end-to-end business processes by consolidating data from multiple interrelated custom and packaged applications and de-identifying confidential information to protect privacy.

Metadata management

* We capture business and technical metadata to establish a common vocabulary between business and IT.

* We collect operational metadata to understand the complete lineage path of information.

* Our metadata management environment supports metadata standards, change control, and configuration management.

* Users have real-time access to our metadata repository from any desktop application.

* We use metadata to perform Impact Analysis (that is, to understand the downstream effects of changes to a data element).

* Our business rules management infrastructure is integrated with our metadata management infrastructure.

Content management

* We use a robust platform to capture, store, manage, integrate, and deliver all forms of content.

* We manage and control the complete content lifecycle, including tasks such as publishing, expiration, and retention.

* We monitor content management system performance in real-time and generate reports (for example, management reports and trend analysis) and alerts.

* Content is componentized, enabling it to be created once and published in multiple formats and channels (for example, online, as hard copy, in book format).

* We federate content from heterogeneous repositories. Trusted Information (TI)

Information integration

* We easily connect to any data source and perform complex data transformations and integrations.

* We capture changed data directly from database logs (rather than from database queries) so that we do not impact the performance of our mission critical applications.

* We have a Gateway strategy and tooling in place that provides access to federated data sources across the enterprise.

* We transform and integrate data in batch mode, in real-time and as a web service as needed.

* We transform and integrate data on a massively scalable parallel platform.

* We have scalable shared services for data flow and data integration across the enterprise.

* We have scalable data flow processing with integrated metadata management and data quality management.

Information quality

* We use data profiling tools to understand the data, validate data values and column/table relationships, and find/ analyze anomalies.

* We enhance information quality using robust standardization, cleansing, matching, and survivorship techniques.

* We reliably match multicultural names.

* We enhance information quality in batch mode, in real-time and as a web service as needed.

* We enhance data quality on a massively scalable parallel platform.

Master data management

* We cleanse and rationalize conflicts in master data to create "golden records and data structures" of master data that is used consistently across the enterprise.

* We can implement master data management in any architectural style (Consolidation, Registry, Coexistence, and Transactional).

* We manage master data for multiple data domains (for example, Party, Product, Account, and Location) and maintain relationships between data domains.

* We have an enterprise master data synchronization hub.

* Our master data management infrastructure is directly integrated with our infrastructure for information integration, metadata, and data quality.

* We maintain an audit trail of changes to master data so we can understand what changed, why it changed, and what master data looked like at any point in time.

* We perform master data management in batch mode, in real-time and as a web service as needed.

Content-centric business process management

* We seamlessly manage, store, and share content that is part of our business processes.

* We automatically respond to business events and transactions as they occur.

* We simulate processes to identify process bottlenecks and optimize processes.

* We monitor processes through dashboards that provide in-depth analysis of live and historical process information.

* We rapidly deploy improved processes while minimizing the impact on operations.

Records management

* We automate records management activities to help enforce compliance with defined retention policies and legal holds.

* Our records management system is DoD 5015.2-certified for electronic records management.

* We manage records in a federated environment.

* We automatically analyze the full text of documents and emails to classify them for retention, archive, or deletion.

* We intelligently archive and deduplicate emails to help reduce storage costs.

Define and Govern (DG)


* We have published an Information Strategy.

* Our Information Strategy is well understood.

* We have published a Plan to achieve our Information Strategy.

* We have policies, organizations, and budgets in place to support our Information Strategy and Plan.

* We have a Business Process Optimization strategy that is tied directly to our Information Strategy.


* We have mature data standards and data models.

* We have a mature information integration architecture.

* We have a mature data warehousing architecture.

* We have a mature business intelligence and analytics architecture.

* We have a mature master data management architecture.

* We have mature data management processes that are used across the enterprise.


* We have an enterprise-wide data governance organization.

* We have identified our information assets and assigned them data stewards.

* Data stewards have clearly defined and well-understood roles and responsibilities.

* Data stewards have strong expertise in both business and technology.

* We have a standard, auditable process for resolving data governance issues.

* Data Stewards manage change control and configuration.


* We define key data elements and maintain these definitions in a metadata repository.

* We standardize data definitions across the enterprise.

* We track data quality issues and aggressively manage them to resolution.

* Data quality issues are resolved at the source.

* We have identified our critical master data and data structures.

* We model key business processes and understand how data problems in those processes can adversely affect us.

* We can track the flow of a data element from entry to its final publishing or use in a business decision.

* We measure the costs of acquiring and maintaining our data.

* We measure the business value of our data.

* We have a formal security/privacy compliance process.

* We have mature processes to discover inappropriate access to or usage of enterprise data.

* We obfuscate restricted data from nonessential employees.

Analytics and Optimization (AO)

Business intelligence and performance management

* We provide a consolidated, accurate view of performance data across the enterprise.

* We provide a complete range of BI capabilities (reporting, analysis, dashboards, scorecards, and alerts) through a single, Service-Oriented Architecture.

* We deliver performance information to those who need it, tailored to each user's information needs.

* We rapidly analyze planning requirements and "what if" scenarios to anticipate the correct course of action.

* We deliver timely, reliable forecasts and plans.

Advanced analytics

* We use Predictive Analytics (for example, regression, logistic regression, neural networks) to predict outcomes (for example, to forecast future sales and costs, to estimate the probability that a transaction is fraudulent, to predict the effectiveness of alternative courses of action).

* Predictive Analytics are cost-effectively embedded in operational systems.

* We use Text Analytics to analyze unstructured text (for example, to process claims, to classify emails for archive or deletion, to identify product defects based on problem reports, to identify trends in customer satisfaction based on call center records).

* We use Identity Analytics to identify relationships between seemingly unrelated entities (people, companies, organizations, and so on) in order to detect fraud, collusion or identity theft.

* We use advanced data flow/event-driven analytics to accelerate and optimize the business.

* We use advanced streaming analytics for real-time, low latency analytics.

* We use massively scalable parallel engine technology to solve analytic business problems.

* We use Operations Research technology to develop optimal plans and schedules that balance competing objectives (for example, customer service level vs. cost) within constraints (for example, limited resources or time).

Source: IBM

(1) IT capabilities have been defined as "combinations of IT-based assets and routines that support business conduct in value-adding ways" [Sambamurthy and Zmud 2000, p. 108].

(2) Specifying the relationship as a quadratic did not prove helpful or worthwhile.

(3) For the sake of brevity, these results are not shown here, but are available on request.


* Suzanne Heller Clain is an Associate Professor of Economics at the Villanova School of Business. Prior to her employment at Villanova, she taught at Bryn Mawr College and worked as a consultant for Ashenfelter and Ashmore. Her research interests include various topics in labor economics and applied microeconomics. She received her B.A. in Mathematics and Economics from Wesleyan University, and her M.A. and Ph.D. in Economics from Princeton University. Matthew J. Liberatore is the John F. Connelly Chair in Management and Director of the Center for Business Analytics at the Villanova School of Business. He previously served as Associate Dean and chair of the Department of Management, taught at Temple University, and held management positions at FMC Corporation. Liberatore has published extensively in the fields of health-care decision making, analytics, supply chain management, operations research, information systems, project management, and R&D management. He received his B.A. in Mathematics from the University of Pennsylvania, and his M.S. and Ph.D. degrees in Operations Research from the Wharton School of the University of Pennsylvania. Bruce Pollack-Johnson is an Associate Professor of Mathematics and Statistics at Villanova University, having also been a professor of Mathematics at Oberlin College. He has published dozens of papers on analytics, project management and scheduling, forecasting, educational modeling, and on teaching applied mathematics, as well as three editions of a two-volume text on business calculus and finite mathematics. He received a B.A. in Sociology from Brandeis University, an M.A. in Applied Mathematics from Temple University, and an M.S. and Ph.D. in Operations Research from the Wharton School of the University of Pennsylvania.
Table 1. Seemingly Unrelated Regression (SURE):
Select Estimation Results for Capability Measures

                            DG:          TI:          AO:
                          equation     equation     equation
                            (1)          (2)          (3)

Constant                  0.832 ***    0.349 **     0.676 ***
                         (0.000)      (0.027)      (0.000)
IF                        0.659 ***    0.679 ***   -0.013
                         (0.000)      (0.000)      (0.816)
DG                           --           --        0.164 ***
TI                           --           --        0.550 ***
Controls for Industry,      Yes          Yes          Yes
  Geography and Year
Adjusted [R.sup.2]        0.418        0.562        0.462
F statistic              18.10 ***    31.59 ***    19.21 ***
N                           382          382          382

Note: The results are generated from the application of
the estimation technique to all three equations simultaneously.
Numbers in parentheses are p-values. *** denotes significance
at the 1 percent level, ** denotes significance at the 5 percent
level, and * denotes significance at the 10 percent level.

Table 2. Logit Estimation Results for Analytics
Investment Decision

Variable                                  Model 1      Model 2

In(Total Revenues)                        0.410 ***    0.435 ***
                                         (0.001)      (0.000)
Current Level of Analytics Capability     0.111 **     0.083 *
                                         (0.030)      (0.073)
Median Industry Net Income Margin %       0.171 **     0.124 **
                                         (0.015)      (0.034)
Company Debt/Equity Ratio (%)            -0.001       -0.001
                                         (0.359)      (0.319)
Constant                                 -4.990 ***   -5.081 ***
                                         (0.000)      (0.000)
Geographical Location Controls              Yes           No
N                                           147          147
% correctly predicted                    72.10%       69.40%

Note: Numbers in parentheses are p-values. *** indicates
significance at the 1 percent level, ** indicates significance at
the 5 percent level, * indicates significance at the 10 percent

Table 3. Results of Regressions Exploring Whether
Capabilities Affect Firm Profitability

                                  Model 1     Model 2     Model 3

Constant                         -16.605     -19.107     -11.273
                                  (0.193)     (0.472)     (0.547)
Median Industry Net Income         2.996 *     3.427 *     3.392 *
Margin %                          (0.065)     (0.092)     (0.092)
Current Level of Analytics          --        -0.253        --
Capability                                    (0.911)
Higher than Industry Mean           --          --       -19.642
Level of Analytics Capability                             (0.288)
Adjusted [R.sup.2]                 0.011       0.005       0.012
F statistic                        3.440 *     1.442       2.014
                                  (0.065)     (0.240)     (0.137)
N                                   216         167         167

Note: Firm net income margin (as a percent of sales)
measures firm profitability. Net income margins are measured
one year after analytics capabilities. Numbers in parentheses
are p-values. * indicates significance at the 10 percent level.
COPYRIGHT 2016 The National Association for Business Economists
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Clain, Suzanne Heller; Liberatore, Matthew J.; Pollack-Johnson, Bruce
Publication:Business Economics
Date:Jan 1, 2016
Previous Article:Secular outlook for global growth: 2015-34.
Next Article:Fed funds rate surprises and financial markets.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters