Enterprise analytics: What IBM learned

Brenda L. Dietrich, Emily C. Plachy, and Maureen F. Norton are the co-authors of Analytics Across the Enterprise: How IBM Realizes Business Value from Big Data and Analytics, a new book that provides an inside look at how IBM derived value from analytics across nine business functions and the lessons the company learned in the process.

The authors share real-world perspectives on what does and doesn’t work and how organizations can start or accelerate their own transformation to data-driven decision making.  They spoke with TDWI’s James E. Powell  about how organizations can effectively leverage big data and business analytics.

What's the biggest mystery enterprises face as they begin their analytics journey?

Plachy: One mystery enterprises face is how to get started with analytics. The answer is to start with your most important problem. Next, ask questions to help refine what you want to learn, then look for data than can help answer the questions. Form a team to prepare the data and create the analytics model. The team should be composed of an experienced data scientist, someone experienced in business, and an IT person skilled in analyzing the data. Work incrementally on the data and iterate on the model.

Norton: Another mystery enterprises face is how to cultivate an analytics culture within an enterprise to use the analytics insights to improve an outcome. If analytics are done for analytics sake, it is "just fun math" and won't drive outcomes. You have to use insights from the analytics to do something different than you otherwise would have done -- for example, changing a strategy, decision, or behavior.

Dietrich: A third mystery involves measuring the impact of analytics, especially when the deployment of analytics is coupled with changes in the business process or flow of information. Applying analytics to historical data, and comparing the impact of the recommended decisions to the outcome of the actual decisions can be useful in building a business case for deploying analytics. The use of a carefully controlled pilot deployment, for example for a single line of business or geography can enable measurement of impact, as some organizations will be taking action based on analytics while others will not.

What are the three biggest mistakes organizations make in their transformation to leverage data and analytics?

Dietrich and Plachy:

1. Failing to prepare for deployment by preparing the analytic methods to serve the end user in his/her job. To realize value, people must use analytics results to drive decisions and actions. Early in the life cycle, target user should buy in to the solution and be ready to use it.

2. Spending time to understand and prepare data without being driven by a business problem; teams assuming that, "If we build it, they will come." Instead, let the business needs drive the order in which data is understood and prepared.

3. Working to create a perfect solution in one step – analytics teams debating among themselves rather than focusing on the end users' needs. Delivering analytics solutions incrementally has several advantages, including helping  users buy in to the solution and allowing people to use increments that provide insight they did not have before, allowing them to make better decisions.

What "Aha!"s or lessons learned came from the case studies in your book?

Norton: The case studies demonstrate that analytics is a way of doing business and not just a technology. Insights from the analytics have to be embedded into existing business processes to have an impact and transformation is itself a process, not a project. One finance person we interviewed summed it up best. When asked about the transformation, the person said, "We will never be done."

Dietrich: Another "Aha!" was that creating a learning, adapting organization requires analytics. The organization has to keep track of what is known (data), how it is interpreted and acted upon (analytics), and both the expected outcome (from the analytics) and the actual outcome (from new data). This new data describing how the organization works, then becomes a part of what is known.

Plachy: A lesson from the book is that it is not necessary to understand how an analytics technology works to get value from it. You do need to learn how to use an analytics solution effectively, but it is not necessary to understand the inner workings.

What did you find most compelling in writing this book?

Norton: What I found most compelling is the breadth of challenges that analytics can help solve. For example, reliable studies indicate that between 50 and 70 percent of merger and acquisition deals fail. IBM developed a Mergers and Acquisition solution to help identify potential acquisitions and support IBM's growth strategy. The analytic solution identified 18 key attributes that are used to assess potential acquisitions. The model develops the information and the expert user adds the subject matter expertise and provides advice for the business. IBM's acquisition portfolio performance is ahead of the industry which contributes directly to the growth strategy.

Plachy: I found the importance of cognitive computing in analyzing vast amounts of data compelling. Cognitive computing is a new form of computing characterized by computing systems which sense, learn, reason and interact with people.

Jeff Jonas, one of our IBM Fellows, has a powerful quote: "The most competitive organizations are going to make sense of what they are observing fast enough to do something about it while they are still observing it." Cognitive computing is coming just in time to allow us to act in time by providing visualization of big data insights based on our questions, by helping us explore data and uncover insights and by helping us detect anomalies – and this is only the beginning.

Who should read your book?

Norton: This book should be read by current and future business leaders and students who want to boost their careers and stay on the leading edge. For current business leaders, the recommendations in the book inspire a can-do approach to getting started or accelerating their journey. Every business has challenges, and the approach outlined shines a light on new problem-solving approaches that are proven – the book contains real-world examples, lessons learned and a way to think about adding more science to decision processes to drive outcomes.

Dietrich: Data scientists and others who work in the field of analytics may also enjoy the book. Although it won't divulge any new algorithms or models, it may give them ideas on new areas to which analytics can be fruitfully applied.

A longer version of this article originally appeared on TDWI, a sister site to GCN.