« Using decision management to transform claims | Main | Growing your business with decision management »

Analysis and creativity in decision-making

Scott Thurm recently wrote a piece in the WSJ Marketplace section - Now, It's Business By Data, but Numbers Still Can't Tell Future. He talked about the growing trend of trying to run companies more analytically, more "by the numbers", and the success some of those companies have had with this approach including those profiled in Tom Davenport's book "Competing on Analytics" (reviewed here). He quoted Robert Sutton (one of the authors of Hard Facts, Dangerous Half-Truths And Total Nonsense, reviewed here) who contrasted running a business by the numbers with running it based on "faith, fear, superstition and mindless imitation"! However, Scott then goes on to identify two key risks inherent in an analytic or data-driven approach.

  • Change upsets the basis for the analysis
  • Too much focus on analysis stifles the creativity needed for innovation and tomorrow's growth

These are both good points and made me think about enterprise decision management(EDM) in this context. Now Tom Davenport talked about the need to make "quick, accurate decisions on an industrialized scale" when he reviewed Smart (Enough) Systems. These kinds of high-volume, operational decisions are the focus for EDM and change to the environment in which you are making those decisions must be considered. No decision can be automated in a way that will ensure it remains effective indefinitely - changes to competitors, markets, products and economic conditions will conspire to ensure it degrades over time. If you are lucky, it might degrade gently. If you are unlucky, some sudden shift could ruin you.

This need to manage and improve decisions in the face of change is why challengers is so important to the successful adoption of decision automation. With adaptive control you constantly test your current decision making approach against challengers to see if any of them work better. This helps both find better approaches and spot when your existing approach is no longer optimal. An infrastructure for adaptive control also allows you to move into true experimental design where you are systematically checking a large number of potential approaches and aiming for continuous optimization.

Some changes are too sudden for this approach, however, as the results of many decisions take a finite length of time to collect. A very rapid change might mean you are in trouble before the results show it. Using decision simulation techniques, and a robust model of what influences the decision, many organizations are now running scenarios (such as much higher interest rates or a bad hurricane season) to come up with the rules and analytic models that work best in those circumstances. These decision approaches can be kept on the shelf ready to go in case one of those major changes should occur. Even this does not completely solve the need to respond to unexpected change, for which a general focus on agile information systems is about all you can do, but it limits the circumstances in which you will have no response beyond "gut feel". Now that I have finished Harry Potter 7 (of which more later), I am also in the middle of reading Nassim Taleb's The Black Swan: The Impact of the Highly Improbable and I will write up a review and some thoughts on managing "black swan" events sometime soon.

Even with adaptive control you run the risk that you are perfecting today's business rather than thinking creatively about tomorrow's possibilities. However, when you are dealing with massive scale (millions of accounts, hundreds of thousands of customers etc), you really have to be able to model the impact of your creativity before you put it into production. Changing a web page layout may be easy enough to try and see if it improves people's user interaction, but introducing new pricing or a new product without having some idea what impact if might have on other products, shelf space, marketing campaigns etc is probably foolish. I would also suggest that creativity and analytics can go hand in hand, as I have discussed before in my review of Malcolm Gladwell's Blink and of Larry Rosenberger's "Future of Analytics" presentation at InterACT last year.

Technorati Tags: , , , , , , , , , , , , , , ,

First time on the EDM blog?
Subscribe to the EDM blog feed or check out some other recent posts:


TrackBack URL for this entry:

Listed below are links to weblogs that reference Analysis and creativity in decision-making:


Mark Eastwood

I think Scott Thurm ( misses the boat. While I agree that if the rules of the game fundamentally change, analytical models can’t help and I believe modelers would agree. That said, so long as this doesn’t happen they perform well. Furthermore, most experienced modeling organizations (such as Fair Isaac) understand that things change over time and recommend that models are updated or rebuilt on a periodic basis.

Scott’s correct when he says “Running a complex enterprise can't be reduced to a spreadsheet, however. Even the most detailed statistical analysis has limitations, as Mr. Sutton acknowledges.” No one says that models are 100% perfect 100% of the time, but they are statistically accurate a very high percentage of the time. Furthermore, it’s unreasonable to model all decisions or implement them in a EDM platform all the time. So again, he’s correct in saying that you can’t run your entire business “off a spreadsheet”, but I think that’s throwing the baby out with the bath water.

There is a place for analytics and EDM and there is a place for human judgment and there’s a place for statistics and models and other inputs to influence human decisions (I’m thinking of DSS - decision support systems).

Scott quotes an interesting observation by Stanford business professor Robert Sutton about innovation. Business can’t be so focused on the perfect analytics/EDM that they forget to innovate. There are many reasons why businesses fail to innovate or change with the environment; its imperative for any business to make time for innovation. I believe EDM can be a basis for addressing the risks addressed by Scott. First, a robust EDM implementation should provide a mechanism for subject-matter experts to make changes to business policies, procedures and decisions “at the speed of business” that is at the speed of environmental and competitive change. Naturally this must include the usual safeguards of testing and review, but that doesn’t mean it can’t be fast and efficient. Models should be constantly reevaluated to ensure they are actually predictive (e.g. a feedback loop) and challenger strategies and models should be part of the system to allow for the mangement of continious change as the environment changes.

The value of automating decisions (I include here rules + models) is too significant to be ignored just because some "black swan" event might occur.

The comments to this entry are closed.

Search Site



  • enter your email

Upcoming Events

  • FICO Tools & Analytics User Forum 2012
    BERLIN: September 11-12, 2012 LONDON: September 18-19, 2012 Gain new insights for improving business performance through advanced analytics and decision management tools.