Holistic Detection and Business Intelligence
by James Taylor
Originally published August 12, 2009
One of my favorite authors is Douglas Adams, and he once wrote a book called Dirk Gently’s Holistic Detective Agency. In it, one of the characters describes an imaginary software product:
You see there have already been several programs written that help you to arrive at decisions by properly ordering and analyzing all the relevant facts so that they point naturally to the right decision. The drawback with these is that the decision which all the properly ordered and analyzed facts points to is not necessarily the one you want.
Well, Gordon’s great insight was to design a program which allowed you to specify in advance what decision you wished it to reach, and only then to give it all the facts. The program’s task, which it was able to accomplish with consummate ease, was simply to construct a plausible series of logical-sounding steps to connect the premises with the conclusion.
As I was re-reading this the other day, it seemed to me that this describes not a product but the typical decision-making process in many businesses – the people making decisions decide what answer they want and then use their business intelligence tools to help justify their decision. Sometimes they use analysis to generate various “facts” that justify the decision. Sometimes they simply make the decision and then use analysis to show that it was, in the end, a good one. Indeed, sometimes it seems to me that whole swaths of executive information systems of one kind or another can be lumped into this “justify my decision” category (though I admit to being a cynic). And perhaps this is not a complete disaster when one is talking about executives and experienced managers. After all, one hopes their experience and judgment have been applied, however implicitly or indirectly, and that the decision will be a good one (or at least a reasonable one).
But if the person making the decision is a junior member of staff, a new hire or a poorly paid front-line worker, this approach is less appealing. We don’t want folks at the front line making decisions and then using their BI tools simply to justify them – we want them to make valid, legal, appropriate and informed decisions. Properly ordering and analyzing all the relevant facts, and helping you determine which facts are relevant, is the role of business intelligence in most organizations. And this is fine for many decisions – decisions that are one-of-a-kind or important enough to justify serious, well-trained, experienced staff.
If these decisions repeat – if we make the same kind of decision over and over again – we also want to know how these decisions were taken. If we just have the justifications people use, then we have nothing to analyze – no way to see how the decision is being made and so no way to try to improve it. Without a record of the steps followed to make the decision, it will not be possible to compare different decision-making approaches to see what worked. Without some explicit decision-making steps to follow, it will not be possible to experiment. And if we don’t understand how people are using analytic insight, or if they are not really using it at all, we will not be able to show an ROI for all our analytic tools and techniques. For repeatable and repeating decisions, then, we need a different approach.
In our book, Smart (Enough) Systems , Neil Raden and I identify operational decisions as those that repeat most often and most consistently. These decisions, while not individually that valuable, add up because companies make them so often. The actions taken as a result of these decisions drive transactions, and better decision making drives value customer by customer, order by order, and offer by offer. Managing these decisions effectively requires decision management (sometimes called enterprise decision management or business decision management).
Decision management automates and improves these operational decisions. Starting with the decision in mind, decision management identifies the data and analytic insight that could contribute to a better decision. It combines these analytics with the policies and regulations – the rules – that drive and constrain the decision, often laying them out in a series of steps.
Each piece of analytic insight, each rule, is managed independently. Not only does this allow them to be managed and reused more effectively, but also it means that each decision made can record precisely which rules, which analytic models, contributed to the decision and how. Ongoing analysis of these decision-making logs – decision analysis as it is called – can map results to the decision-making process used to see what works and what does not.
Decision management also emphasizes experimentation, so the ability to try different decision-making approaches on different customers or different offers is critical. The use of adaptive control, testing different challenger approaches against an established champion to see which works best, creates an environment that constantly tests and learns.
These elements – a focus on operational decisions, the integration of analytics into them, the explicit management of decision-making rules, ongoing decision analysis and the integration of adaptive control – are what define decision management. Decision management creates the environment for making sure you get the decisions your data points to and that your policies require.
Recent articles by James Taylor
Copyright 2004 — 2019. Powell Media, LLC. All rights reserved.
BeyeNETWORK™ is a trademark of Powell Media, LLC