(Posted by Guest Blogger, and BI Skeptic, Ian Turvill.)
I'm working in London today, ahead of next week's InterACT Conference in Lisbon, and I've been so busy juggling responsibilities across time zones that, first of all, I deleted my daily CIO Insight email from my inbox. But when I considered the title of one of the stories listed in it (Business Intelligence: What Are You Thinking?), I knew that James would never forgive me for passing up the opportunity to blog on it.
"Por que?", you ask. (That's Portuguese.)
Well, let's see: because James has written extensively on the shortcomings of Business Intelligence, and why organizations should consider not just the insights they garner from their data, but also how they can convert those insights into action through Enterprise Decision Management.
When I clicked on the article, I saw that the author had much the same opinion:
"I believe routine business decisions should be as automated as possible. After all, when a decision is routine and you understand the rules well enough, that decision can be made by software—using a decision engine—acting on those rules in a consistent, unbiased and auditable fashion. We've seen a lot of examples of this over the past few years, such as program trading in securities, granting consumer credit, mortgage origination, inventory management, and support-desk escalation processes."
Amen, brother! We're singing from the same hymn book!
Plainly, however, the author is not a regular reader of our blog, because he continues:
"Because this is a new category, these tools tend to be a collection of not-yet-very-well-connected ideas from different sources."
Obviously, while James and I wouldn't dispute that there are many different sources, but we hope at least that no one would call our thinking "not-yet-very-well-connected".
So, here are some immediate suggestions that perhaps the author, John Parkinson, could add to his "thinking" repertoire.
- Enterprise Decision Management (expanding on the idea of decision engines) is a systematic approach to automating and improving decisions. Through business rules management technology and advanced predictive analytics, EDM can give your most critical, high-volume, operational decisions greater precision, consistency and agility.
What's missing from Mr. Parkinson's treatment is any sense of analytics. Decision automation has to combined rules AND analytics; otherwise, the sophistication of decision making is just insufficient.
Here's another "thought".
To determine what constitutes a “good” decision process, it is critical that executives understand the many different facets of a decision that contribute to overall business performance.
This holistic way of evaluating decisions is what we call "Decision Yield".
Decision Yield is designed to evaluate automated decisions which are typically: customer-facing—from approving loans to pricing insurance to determining cross-sell offers; very frequent—often many thousands of times a day; driven by strategies and rules, and often supported by predictive models, decision models and optimization; executed in real-time (credit overlimit approval) or in batch mode (matching an offer with a prospect).
John emphasizes that decision engines should be able to act on chosen strategies in "a consistent, unbiased and auditable fashion." The concept of Decision Yield suggests a broader set of measures are needed. Consistency is certainly important, but precision (making the right decision), agility (changing decision strategies quickly), and the speed and cost of decision making are also very important.
There's obviously a lot more that could be said on this topic, much of which is distributed through this blog. We certainly invite John to take a look and contribute his thoughts about how this category might develop. Unfortunately, there doesn't seem to be an opportunity to offer comments on the CIO Insight site: Perhaps a quick email to the editors will suffice.
Important caveat to this article:
It's not ENTIRELY clear to me from Mr. Parkinson's article, if he is conflating the concept of decision automation, which revolve around embedding decision-making into operational systems, with the idea of decision support, which centers on creating the strategies that the operational systems will apply; or if they are instead entirely different ideas for him.
Perhaps I need a "thinking machine" to aid my comprehension?
In any event, I think the points I make above still hold true, because EDM really applies to both needs: the offline development of strategies in a flexible, analyst-friendly environment, and then inline application of strategies in the very many application types John mentions.