« Rules and requirements - what do YOU think? | Main | Decisions and the potential for agile services »

Data Mining, Predictive Analytics and Markets of 1

Two articles made me think about the application of enterprise decision management, EDM, in the world of customer experience. First, I saw the one Jeff Kaplan wrote - "Data Mining as a Service: The Prediction is Not in the Box". Jeff asked the relevant question very early on:

"Why were there so many failed enterprise customer relationship management (CRM) implementations? "

and he went on to suggest that the use of predictive analytics (see my FAQ) should be pat of the solution. He then assets that:

"The key challenge that packaged predictive analytics software has not been able to crack is how to extract knowledge from data quickly and put it into the hands of marketers to make better, more informed decisions"

Here I have to disagree with him slightly - it is just not about putting knowledge in the hands of marketers so they can make better decisions, it is about putting "knowledge" into those failed CRM implementations so that the systems can make better decisions. This requires more than just a focus on predictive analytics - it requires a focus on automating, managing and improving decisions or EDM. Jeff's outline for applying predictive analytics is a good one, however, in particular as he correctly identifies need to focus on the problem first and then gather and analyze the data (as distinct from the approach often taken of collecting data and hoping something can be discovered from it). I would add a couple of things to his outline:

  • You need to focus on the decisions you want to improve - cross-sell offer, home page design, call routing or whatever - as this gives you the problem statement you need
  • You need business rules not just analytics. There are rules derived from policy, from regulations, from experience or even managed directly by customers that must be considered. A decision is a great place to gring the rules and analytics you need together.
  • When you get into the automation of decisions you need adaptive control to manage your experimentation and testing and to ensure that you can act on the results you get. This boils down to a software architecture to let you do in your system what your analysts would do in their heads or with Excel - compare approaches, analyze them for success, continually improve.

You should read the article as it makes some great points about predictive analytics and the challenges in using them, especially for the first time. Moving on, the second article was by Rob Walker of Chordiant - Next-Best-Action Marketing: Creating the Segment of One. My favorite quote from this one was:

"combine predictive models that were once the province of statisticians with a completely new breed of user-oriented business rules that can significantly improve the customer interaction experience"

Wow - sounds like EDM to me :-) Actually my friends at Chordiant call it "Chordiant Decision Management" and are focused more singularly on customer experience decisions but hey, close enough. I have already blogged this week about using EDM to meet CRM challenges and use what you know about your customers and interestingly enough Analytics + CRM = Happiness was one of my first posts on the blog - clearly this is a topic that comes up a lot. Rob's article was very informative, and well worth a read, as he (like me) is focused on a "corporate decisioning hub" for delivering decisions hither and yon (check out this article for my POV as written up in The business rules revolution). Building a customer intent driven organizations takes this kind of ruthless focus on customer treatment decisions as being customer-centric means being decision-centric. Such a focus allows you to deliver extreme personalization and markets of 1 and 1:1 communication with scale. Using analytics to segment customers and rules and more analytics to target them very precisely helps increase customer loyalty and perhaps recreate the feel of "the corner store". It also allows you to survive in a Long Tail world with both hits and niches. I have written before about both EDM and Customer Centricity and The "Best Next Action" as well as about the experience of a Fair Isaac customer - Customer Centricity in Action: Best Buy.

If you don't want to read all those blog postings, you could buy and read Smart (Enough) Systems - the book I wrote with Neil Raden - as this is exactly the kind of situation that demands smarter (CRM) systems. Other books you might enjoy on this topic include Chocolates on the Pillow Aren't Enough (a great book on customer service), Competing on Analytics (a good introduction to the power of being an analytic competitor), Berry and Linoff's classic Data Mining Techniques and (for those of you focused on lots of niches for your products). The Long Tail.

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 Data Mining, Predictive Analytics and Markets of 1:

» Data Minining, Predictive Analytics and the Market of One from The Perfect Customer Experience
Good article on how analytics and business rules lead to more effective customer experience marketing at Enterprise Decision Making blog by James Taylor, with good references to other similar articles. For example:Building a customer intent driven orga... [Read More]

» Who Really Wants Predictive Analytics? from BI Questions Blog
The value of predictive analytics is obvious: who wants to "drive looking out of the rear view mirror"? But in practice, predictive analytics hasn't been widely implemented. What might change in the future? (more...) [Read More]


Jeff Kaplan

JT – thanks for forwarding on your blog link...I appreciate your write-up. Your points are extremely valid in particular the use of incorporating logic/rules and automation as this is part of our standard practice. The reality is as I’m sure you see this yourself with many businesses is that they’re still in the process of taking the leap in investing in predictive analytic we try to boil up the high-level points in an effort to first educate and develop an ROI case...with the end goal of implementing a living breathing model(s) that is continuously learning and can be measured (to your reference of test and control).

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.