Responding to the different needs of customers begins with identifying different behavioral groups--but all database segmentation means to most companies is more targeted selling.

When asked about her company's approach to database segmentation, a Market Research Manager was emphatic in her response:

I'm not a fan of behavior-based segmentation. I don't want to say it's a waste of time, but I think people waste a lot of time on it. I think it really makes for good bed-time reading, but there's not a whole lot you can do to action it. You can't just put a label on someone, append it to their database record, and assume they'll always behave a certain way.

Considering her company had a data-rich transaction database of over one million customers, it was a perplexing comment. And she was not alone in her opinion--even the various product managers at the company viewed the database as simply a convenient list of names to target. According to them, the database was made up of prospective buyers, not customers with a need for multiple products. No one even could even guess the average cross-sell ratio – because no one had ever cared to know.

Segmentation is supposed to be the cornerstone of CRM. By grouping customers into discrete sub-groups based on how different (or similar) they are in their characteristics and behavior--or so the theory goes--new products and offers can be dreamed up that will appeal to their distinct needs. In practice, most companies simply partition their customer bases into target groups for direct marketing purposes. A segment, in that narrow context, is merely a group of customers with a strong likelihood to purchase.

Even when a company does try a more strategic approach, statistically segmenting customers into unique behavioral groups, it does not automatically lead to a switch in thinking. Take one major department store retailer which some years ago hired an external agency at great expense to conduct a cluster analysis of its multimillion-name database. The store made some intriguing discoveries--such as realizing that one-fifth of its customer file habitually returned products, while an avaricious 2% bought anything and everything--but the results were greeted by management with a shrug of the shoulders. More interested in what was being sold than who was doing the buying, the retailer had no clue how to apply the insights it had gained--or much interest in doing so. The cluster findings soon faded into memory while the company carried on with its mass promotional mailings.

A similar case involves a financial services provider that recently completed a cluster analysis of its customer file. The results were not particularly startling: 10% of the database consisted of older long-tenure customers who preferred to make their deposits and withdrawals in-branch, while 35% were single-account holders who generated hardly any profit at all. But they did provide some clue as to the unrealized potential that existed within certain segments.

Nevertheless, the institution is now bogged down figuring out what to do next, uncertain how to shape its marketing plans and programs around its key segments. For now, the segment profiles remain simply “good bed-time reading.”

So despite bigger and better data warehouses, fancier analytical tools, and a growing body of data-mining knowledge, companies continue to struggle with the concept of database segmentation. A major reason for that is the way marketing decisions get made: power resides with the product managers who have little interest in treating customers other than as a target group for their own offerings. The most they might collectively agree to do is co-bundling, usually in order to generate sales on the back of a companion product. Unless companies enforce discipline on the CRM process, the job of meeting sales targets will always override the best interests of the customer.

Another major factor is that omnibus segmentation schemes, such as the kind produced by multivariate analysis of a database, are too layered for most marketers to make sense of--and often too volatile for long-term planning. Segments whose determinant traits are formed out of many base dimensions--product affinities, demographics, credit history, touchpoint interactions, channel preferences, Web usage and more--are prone to abrupt transformation, even complete dispersion, as a result of marketplace events. All of a sudden an attractive segment that might have been the focus of a new product initiative is nowhere to be found, the customers subsumed by neighboring segments or too different to be recognized at all.

That's why companies prefer to deploy a variety of segmentation methods to suit different uses: attitudinal clusters for brand development, RFM cells for direct marketing, and geo-demographic segments for market analysis. But working with all of them at once is like toying with Rubik's Cube: it can be frustrating getting them to fit neatly together. For instance, attitudinal profiles derived from a stratified sample of the population are impossible to apply with any confidence to database records. So while application-specific segmentation may ease comprehension, it doesn't answer the need for a broad scale solution that simultaneously takes into account how customers look, act, and think.

The trouble with multi-purpose database segmentation is that it is hard work. Segments have to be sharply defined, remain stable over time, and be deserving of separate attention. To pull it off, marketers should bear in mind these guidelines:

  1. Start with Who Matters Most: Rank customers by current value and stratify them into tiers. That sets up the most critical axis of the segmentation matrix. The next step is to determine the distribution of high, medium and low value customers across the explanatory dimensions of behavior, lifestage and lifecycle.
  2. Less is More: Choose segmentation variables that appear to be the most logical discriminators. Relegate all other data to a descriptive role.
  3. Find the Magic Intersections: Pick out those cells in the matrix that are meaningful in both size and value; drill down if necessary to discover otherwise undetectable nested sub-groups.
  4. Raise Staff Consciousness: Keep the number of segments to a manageable number and build easy-to-grasp profiles for company-wide distribution, along with a nomenclature that embodies the character of each segment.
  5. Close the Gaps: Use research to uncover the latent needs of the different segments. Those discoveries ought to drive customer strategy.

Identifying the major differences between customers--and grouping them accordingly--will surely lead to development of more sought-after products and services. And that is hardly a waste of anybody's time.

Subscribe's free!

MarketingProfs provides thousands of marketing resources, entirely free!

Simply subscribe to our newsletter and get instant access to how-to articles, guides, webinars and more for nada, nothing, zip, zilch, on the house...delivered right to your inbox! MarketingProfs is the largest marketing community in the world, and we are here to help you be a better marketer.

Already a member? Sign in now.

Sign in with your preferred account, below.

Did you like this article?
Know someone who would enjoy it too? Share with your friends, free of charge, no sign up required! Simply share this link, and they will get instant access…
  • Copy Link

  • Email

  • Twitter

  • Facebook

  • Pinterest

  • Linkedin


Stephen Shaw is vice-president of strategic services with The Kenna Group, a full-service customer relationship management company. He can be reached at 905-361-4046 or via email: