Over the past several months, numerous business people have asked about the salience of the segmentation approaches offered by a number of companies. These offerings are generally base upon demographic information available from a variety of sources especially the government census and its projections. The segments often come with cute names similar to those we have heard over the years like “soccer moms”.

Our response is that one must be careful in employing these pre-designed segmentation approaches because they are not based upon things that are related to the business at hand. Rather, they are generic. As with “soccer moms”, the capacity of the generic, pre-designed segmentation is limited. In 1996, “soccer moms” were defined as women with young children of pre-school or early school age. These women were often imagined to drive an SUV and cart their children to all sorts of events. Moreover, they were going to decide the 1996 election. As it turned out, women that fit the definition of “soccer moms” voted almost exactly as women writ large who voted for Bill Clinton 53% to 39%. So, while interesting for the media and the political junkies to talk about, there was no differentiation provided. Political party affiliation, family party affiliation, group membership and previous voting behavior determined present voting behavior. Behavioral segmentation would have been far more powerful in 1996 than the demographic definition. However, it was also far less interesting.

So it has proven in the business world as well. We have tested a number of segmentation approaches defined using demographic and attitudes. The objective has been determining the salience of the segmentation in predicting behavior. In each instance, other characteristics were far more salient in predicting behavior than these approaches. Moreover, in most cases, the segments overlapped enough so that they really did not exist as independent groupings. That is one reason why you see coupons dispensed at grocery and other stored based on behavior. In addition, segments differ substantially across product categories so a one size fits all does not work.

How individuals have behaved and been treated in the past is a far more accurate predictor of future behavior than demographic segments with attractive names. Individuals who purchase a particular product consistently continue to purchase that product unless something powerful intervenes. New customers need to be reinforced in their buying decision and previous success in certain activities by the seller is the most powerful determinant of future success with new customers. So, while demographic segmentation often offers catchy descriptions of segments, they seldom work properly. Moreover, these demographic segments often overlap with one another further limiting their power.

Segmentation’s real power often comes from utilizing a company’s own data. This data provides behavioral insights into the customer providing critical strategic and tactical information often unused or underused. Powerful segmentation approaches meet three criteria. Segments must be:

1. relevant;
2. independent;
3. predictive.

It is our view that if the segmentation approach fails these criteria, it should be discarded.