Decisions, decision management and analytics

Tom Davenport was interview recently by the Sloan Business Review on Reengineering your decision making processes about analytics and how companies make decisions. While the interview is mostly focused on manual decision making, many of the points are just as valid when you consider decision management and decisioning technology as I do.

Analytics, says Tom, are predictive and explanatory. They focus on the future and on explaining what the data you have collected means not just on reporting what that data is. I have blogged about analytics and what it means before but I always come back to my favorite phrase:

Analytics simplify data to amplify their meaning

As Tom says, analytics take your data and tell you what it means now and in the future, analytics help you see patterns that explain your business and how it’s going. Yet, no matter how sophisticated your analytics are, they won’t necessarily improve your decision making. As Tom points out, you must tie these analytics to actual decisions and make them part of your decision making process (whether manual or automated).

I call this beginning with the decision in mind: Rather than starting with the data and seeing how it can be analyzed, begin with the decision you wish to improve. Understand how this decision affects your business (what KPIs it impacts for instance) so you can understand what makes a good decision and what it means to make better decisions. Then figure out what analytics would help make better decisions and go find, clean and integrate the data you need for these analytics.

Tom also discusses the historical separation between transactional and analytic / decisioning systems. This separation was something Neil Raden and I discussed in Smart (Enough) Systems in the chapter “Why aren’t my systems smart enough already?” Driven largely if not completely by technical rather than business or logic considerations, this separation is finally going away. And it really needs to – businesses are run using transactional systems and if these systems can’t make or support decision making then inconsistent, judgmental and inaccurate decisions may well be the result. Adding Decision Services to legacy “dumb” applications bridges this separation without requiring complete reengineering and makes these systems “smarter” and more analytical.

Finally Tom reiterates something that he and I have long bemoaned – the lack of any systematic attempt by most companies to identify the decisions that matter and to focus their analytic effort on those decisions. Rather, most companies are opportunistic – applying analytics and other decisioning technologies and approaches as and when projects come up. As Tom points out, we need a way to help companies adopt analytics systematically even when they are not headed by someone with an analytic or technical background. I have blogged about this before and I really like Tom’s suggestion is to identify the top 5 strategic and top 5 operational decisions (see this post for a discussion of the difference).

When it comes to operational decisions, I like to suggest that people begin by identifying the decisions within a business process or a set of business processes. The mapping of these decisions to the Key Performance Indicators for a company or even a division can be very enlightening, quickly identifying a set of operational decisions that have a real impact on those things the company cares to measure. This Decision Discovery is the first step of the decision management methodology that Neil and I described in the book and that I have been developing as I have subsequently worked with various clients. The end result is to make operational decision making a corporate asset.

Tom’s new book (Analytics at work, reviewed here) is highly recommended and, interestingly, some research I did with IBM resulted in a very similar pattern of adoption.

Syndicated from International Institute for Analytics