Tackling Common Analytics Problems
By Stephan Kudyba, Apr 07, 2015
So you’ve jumped on the analytics bandwagon to improve decision making and process efficiency with strategic information, but so far your plans are not working out…what’s the problem you ask?
As powerful as analytics can be to provide decision support in organizations, there are a number of factors that can render analytic initiatives lackluster or of little value. At a high level, these can involve:
- The gap problem that exists between analysts, IT departments and end users
- Misidentifying the problem to be analyzed
- Not closing the loop in analyzing a problem
- Inappropriate analytic method applied to the problem
- Organizational Issues
- Lack of organizational “buy in” to analytics as a viable resource to make decisions
- Lack of end-user understanding
- Analytics don’t belong everywhere
- Data Issues (Garbage in, Garbage out)
- Sub-par data management tactics
- Lack of essential data variables
- Inconsistent data sources
To maintain brevity for this post we’ll just highlight the major issues that comprise the factors just mentioned.
The Gap Problem
As intriguing as analytics appear…even the best, most complex techniques can provide little value to companies if the results are not addressing the true problem or process of interest to end users. In other words, the gap problems between the analytic process and business management needs is an obstacle that must be overcome. KPIs have to be clearly defined as well as particular process attributes. If BI experts roll out an OLAP cube, dashboard or simple report that doesn’t accurately or adequately describe a process of interest, end users will quickly turn away. Sometimes a report can answer a portion of a CRM, Supply Chain or Customer Service scenario only to get end users asking…but what about this view of the problem over this time, or this scenario. Analytics fail because they don’t provide solid decision support by not closing the loop for users.
Another factor organizations must consider regarding lackluster analytic results is the idea of applying the wrong analytic methodology to a scenario which again, doesn’t provide users with the information they are looking for. A prime example would be the incorporation of retrospective techniques (reports, OLAP cubes) when a predictive, data mining approach, is needed to facilitate “what if” capabilities or forecasting to attain expected or future KPI estimates. Finally, managers need to realize that analytics don’t fit in every area of an organization. Beware of driving analytic initiatives where innovation is the focal point.
Another major scenario that has to be in place in order for analytics to provide value to organizational performance entails the existence of an accommodative culture towards analytics as a viable resource for the decision making process. The potential “sting” or “fear” of analytics has to be dispelled, where end-users’ understanding of analytic roll outs, along with what analytics imply for decision making has to be clearly articulated. Scary “black box algorithms”, complex statistical jargon has to be converted to a business friendly concept, and the idea that analytics are a decision support component that do not dictate decisions but rather offer another frame of informational resource for decision making must be effectively communicated. It is really up to senior management to address these factors to make the transition to a more analytic decision based culture more successful.
The core building blocks to information generation needs to have solid integrity. Data professionals and analysts must take those precautions in eliminating errors in data at the detailed level, removing outliers and addressing unbalanced distributions from the array of data resources they manage. Inconsistent data from duplicate data sources (e.g. legacy systems, data warehouses, and sources existing on PCs) must be identified. Errors, anomalies and odd categorical balances can skew results and project unrealistic or even inaccurate scenarios. This can be detrimental to an analytics endeavor in the event that decision makers suffer catastrophic consequences by relying on garbage models.
Another element in the data integrity concept that is particularly prevalent in the big data environment is the omission of important, “new” data variables that are becoming essential to analytic initiatives. Perhaps end users realize the existence of unstructured sources that may provide an essential decision support element to process information. Data professionals are increasing under the gun to stay on top of the latest data that is available to add descriptive value to processes which can include structured or unstructured, internal or external sources to an organization.
Data resources are quickly evolving, where more sophisticated skills and techniques are required to harness this resource. Even with the best analytic skills and resources, don’t assume analytics will achieve success in your organization. If your initiatives fall short of your expectations, the problems can stem from a number of areas.
About the author
Stephan Kudyba is founder of the analytic solutions company, Null Sigma, which focuses on providing strategic analytic solutions for organizations across industry sectors. He is also a professor in the management department at New Jersey Institute of Technology where he teaches courses that address the utilization of IT, advanced quantitative methods, business intelligence, and information and knowledge management to enhance organizational efficiency. He has published numerous books, journal articles and magazine articles on strategic utilization of data, IT and analytics to enhance organizational and macro productivity.
Dr. Kudyba began his career designing models to trade financial markets in the investment banking industry and later turned his focus to Business Intelligence and Data Mining applications at Cognos Corporation. His latest book, Big Data, Mining and Analytics: Components of Strategic Decisions addresses the critical topic of leveraging evolving data resources to generate actionable information for insightful decision making. Dr. Kudyba holds an MBA from Lehigh University and PhD in economics from Rensselaer Polytechnic Institute.