Segments are useless.

And here's why:

1. Customers are continuous, not discrete

In any customer base, customer purchases are a spectrum. In fashion, for example, the first segment used is often gender, but shoppers buy clothing for themselves and their spouse. In pet food, dog households sometimes get a cat, and it can take months to get resegmented.

How you think your customers behave...

Customer Segments
Actual Customers

... versus how they actually are

2. Segments are hard to target

Beyond email campaigns, it's hard to action any given segment with broad reach, promotions, or price advertising.
How-Advertisers-Think-It-Works
How-It-Works

3. Why be right-ish when you could be right?

It's often just as easy to use the underlying raw attributes instead of the segments. So rather than talk to the "Healthy Nutrition Mom" segment, why not just choose anyone who has bought organics in the past few purchases? It's more complete, and doesn't rely on shaky correlations of other factors?

Penfield implements a "Segments of One" approach which skips the counterproductive clustering and segmentation processes. We score the entire customer master on any number of individual customer attributes, which allows robust and precise analytics and marketing.