Potential: Revealed

Strategic Thinking, Innovative Ideas, Growth Marketing, and Revealing of Potential

Archive for July, 2008

Smart People on Analytics and Segmentation

Recently I asked my network of friends and former colleagues about some work I am doing for a financial services client (“a top 10 U.S. bank”). On the surface they asked me to look at their approach to segmenting their commercial client base and recommend improvements or a different framework. Based on what I found, I essentially asked my network the following:

“Do you concur with my recommendation: the bank has done a good-enough segmentation of their client base but results have been poor due to poor execution. I have told them an increased focus on improving execution will yield better results and is a better investment than seeking an ‘improved segmentation'”.

To share my wealth, I got many good responses and you can see them below. Feel free to respond to any of them and where I could, I provided links to their web sites, blogs or LinkedIn pages.

B2B is usually more stable than B2C, so the data gathered for the segmentation can still be used for future segmentation, if necessary. Segmentation (or Multivariate analysis) can be easily obtained by several different purchasing behavior metrics, or simply by addressing different purchase intent metrics (as well as awareness of services, need gap analysis, current service provider, and switching intention) for prospects. If they are just profiling their house file, and they need to validate their segmentation exercise, they should have held out a sample for control. Maybe the execution was not as bad as the context. It is recommended that a hold-out sample is put aside and no execution is applied to it to gauge what the actual lift is.
So I would combine your approach with a holdout (or control) to run the test (executions) at different levels. Cheers, Roger Ares
Sounds like you’re doing well and are involved in some interesting work with your client. On the surface, your premise does seem reasonable. We have a predictive modeling focus, and we frequently work with clients to develop segmentation strategies, and would concur with your observed delimma of operationalizing and getting expected results from good statistical modeling. Rick Nichols

Doesn’t “Good to Great” or one of those McKinsey books have lots of examples of how even a mediocre strategy well executed does better than a spectacular strategy implemented poorly? I think your intuition is correct; although there is a wide variety of segmentation approaches. A good segmentation doesn’t address all needs (which in itself is counter-intuitive). A good segmentation approaches 1-3 business objectives and optimizes the segmentation against them. The key question is what is the primary objective right now (and the answer isn’t grow sales). It might be lower churn, add users, increase usage, etc. A good segmentation will appropriately focus on the dependent variables that accurately predict that independent variable. Too many dependent variables (the objective) and the segmentation gets diluted.

On some previous work, I had the opportunity to work with a Brazilian beer company to implement a segmentation. The company grew revenues dramatically. While I wish we could take all the credit, I do believe that the good segmentation and the focus on implementing it was a big contributor. And it was hard. There was always pressure to do other things that had nothing to do with the target segments. While they were exciting, we worked hard to only do those things geared toward the target segments. Hope all is well with you, Matthew Hull

I agree that execution is more important that perfect segmentation. If we take the example of a simple organization that focusses on one type of customer only, there is no need for segmentation. When the customer base is more complex than that, then we use segmentation to group similar categories of customers so that we can address their needs, prioritize, etc. So HOW we segment is not as important as WHAT we do after the segmentation. Thanks, Mohan

We find at our company that reaching our clients (new and prospective) in the B2B area is very touch in go. For example, we delivered an IPOD containing our message to100 clients and saw our best ever response and sigh-up rate…yet a month later…we tried it again using the itouch (different audience) and saw half the success. Iteration and experimenting is key. Bill Zielke, VP Marketing, BillMeLater
My two cents…. you are such a breath of fresh air! I agree with the underlying premise that without strong execution, even a great segmentation will fail. You already have a workable segmentation, I would focus on outling the execution strategy and how to get your critical players on board and turning the ‘flywheel’. (did I just say ‘flywheel’? 🙂

There are many examples of success that are a result of a simply ‘mediocre’ foundation, executed flawlessly…… Kellie Brody, Sales Excellence
I generally agree with your thinking. My personal observation is that the evaluation of results from segmentation testing are not detailed enough and or the implementation is not taken to a detailed enough level either for budgetary or “know how’ reasons. Unless the evaulation uncovers a silver bullet, so to speak, the lift will initially be modest at best. What is required is the follow through execution to dig deeper into the reasons for certain behaviors, followed up by using those learnings to further capitalize on opportunities to trigger the desired client behaviors. And so on to zero in on the exact triggers that will change the behaviors in the way that produce the desired lift in results. Best regards, Deborah J. Ackerman

You are correct, if they did segmentation properly but failed at execution then they need to look at what steps they missed. That being said if the segmentation is old (over 18 months then you do need to segment again). Many times financial companies look at a few buckets about customers, whose buying what, from where – when they should be looking at who has the highest margins by customer type and concentrate in going after higher margin customers and drop the volumes of ones that have low margin – same can be applied to products and services.

Dalia Quinones-Zayas
Traditional Market Segmentation at a large bank can be little more than a waste of time in this market. You need deep credit, risk profile and behavioral analytics in order to determine profiles
that can accurately predict lift and better yet the development and expansion of profitable relationships. Think of it like one-to-one but with SOA-like reusable products and services so that you can both identify and execute against patterns within segments and sub-segments quickly and effectively.

Regards, Mike Rouse


Recently I read an excellent blog by author Neil Steinberg called Pursuing Perfection which got me thinking about this concept (yes, concept) called perfection.

Neil gives examples of where “the perfect is not always so perfect”, such as Michael Jordan failing in the final seconds of a 1995 NBA playoff game, or a sphere of pure silicon created during a space flight that upon minutely close inspection has tiny (i.e., 35 nanometer) fluctuations. Small imperfection, yes. Imperceptible to the human eye. But a flaw nonetheless.

In addition he notes that handing out of a judgment that perfection has been attained – for example gymnast Nadia Comaneci earning the first 10’s in Olympic gymnastics history in 1976 – can be momentarily satisfying but fleetingly so. Since that day in 1976, scores of 10 are both more commonplace and questionably the mark of gymnastic perfection.

So, as I like to think about, write about, consult about “revealing and attaining your potential”, I wonder if perfection is the destination, or the journey? In light of the thought that arriving at perfection somehow then ruins it, I’m inclined to choose the journey.