When I interviewed Jeanne Harris and Robert Morison, two of the authors of "Analytics at Work: Smarter Decisions, Better Results," they told me this was due to several factors, including that managers traditionally rely on their intuition and feel overwhelmed by large amounts of data. A recent survey by IT services firm Avanade confirmed that executives increasingly feel they're buried in piles of data, much of it irrelevant.
And, Morison told me, "Most of those systems go in without much forethought [about] how they're going to use this information for decision-making." As crazy as that sounds, I suspect it's true.
I remembered our discussion as I read an International Institute for Analytics research brief written by Niel Nickolaisen, vice president of Strategy and Innovation at Energy Solutions, Inc., co-author of an Addison Wesley book on agile leadership, and a founder of Accelinnova, a think tank focused on improving organizational and IT agility. (The brief is available to members of IIA.)
When Nickolaisen was hired by a struggling specialty retailer to revamp its IT system, he concluded the system's single largest failing was that it provided no decision support. The result: Instead of basing their decisions on data analysis, organizational leaders simply guessed what products to sell and how to sell them.
IT had become a scapegoat for the company's problems, a common reaction. Yet the real problem originated from not establishing upfront the business objectives the company wanted to achieve with technology.
Nickolaisen opted to address this by asking the company's leaders some simple questions that somehow often get neglected in BI initiatives: What decisions do you want to make? What information is needed to make those decisions? (This is the kind of forethought that Morison told me many companies lack.) Then and only then, Nickolaisen writes in the brief, did he begin selecting and developing systems that would generate and make available the data needed to support such decisions.
The new approach yielded several benefits, such as allowing the retailer to replace costly and sometimes ineffective marketing campaigns with less expensive and more targeted direct marketing. The retailer was also able to create "success indicators" that helped it determine which products should receive additional investment to accelerate growth and which should be scaled back or retired.
Focusing first on business objectives rather than underlying technology meant less haggling over application functionality. Nickolaisen writes:
Rather than fighting battles about how to configure the inventory management module of the new enterprise resource planning system, we could satisfy people by confirming that a standard, vanilla configuration would allow them to make all of the decisions about inventory they ever wanted to make.
This project wasn't easy, Nickolaisen writes, because he had to overcome the retailer's ingrained bias toward thinking in terms of transactions rather than decisions. How did he do it? He offers four great pieces of advice:
- Take iterative steps. Enlist the group that is most willing to try and support new processes and/or the group that is in the most pain to begin with targeted prototypes and pilot programs. Again, start with what decisions the group wants to make and the information needed to support those decisions. Select a subset of those decisions as a test case and design processes and implement tools to support them. When improvement is seen, move on to another subset of decisions.
- Envision a perfect future. Base system design on the assumption that, at some point in the future, both technology and access to information won't be limited. Nickolaisen says the retailer assumed they knew they'd be able to one day know when return customers entered their stores and began discussing processes they could implement if this ever happened. Now that smartphones, social networks and location-based technologies are beginning to make this possible, the retailer is nicely positioned to profit.
- Reduce reports. Most companies generate too many irrelevant reports. Nickolaisen advises making sure each report serves a purpose, again by using the "what decisions do you want to make?" question. If a report doesn't support those decisions, eliminate it. He says he's reduced report libraries by as much as 80 percent using this approach.
- Put BI into the hands of decision makers. Nickolaisen's goal is to create as little distance as possible between decision makers and BI tools. So he makes ease of use one of his primary criteria for selecting BI tools. "The tools must be consumable and understandable by someone without a technology background," he says.