Customer Segmentation

What Is Customer Segmentation?

Customer segmentation divides a customer base into groups that share attributes — demographics, behaviors, needs, or value — so that marketing, product, and experimentation can be tailored to each. Good segmentation is the foundation of personalization, targeted experiments, and differentiated lifecycle management. Bad segmentation (too many groups, inactionable dimensions, descriptive-but-not-predictive) clutters dashboards without changing decisions.

Also Known As

• Marketing team: "audience segmentation," "target segmentation"
• Sales team: "ICP tiers," "account segmentation"
• Growth team: "user segments," "cohort definitions"
• Data team: "clustering analysis," "segmentation model"
• Finance team: "value-tier segmentation"
• Product team: "persona-based segmentation," "behavioral cohorts"

How It Works

You analyze 12 months of behavior for 500,000 customers. Clustering on session frequency, feature usage depth, and purchase recency reveals 4 natural groups: (1) 60,000 "power users" — high frequency, deep feature use, $1,200 LTV; (2) 180,000 "steady users" — moderate on both, $400 LTV; (3) 200,000 "explorers" — high visits, low purchases, $80 LTV; (4) 60,000 "dormant" — low on everything, $30 LTV. Now your lifecycle team can run different experiments per segment: upsell for power users, activation for explorers, win-back for dormant.

Best Practices

• Segment on behavior predictive of future value, not just demographics.
• Require each segment to be identifiable in real time, targetable with different experiences, and measurable.
• Pre-register segments in experiment analysis plans to check heterogeneous treatment effects.
• Limit segments to the number the organization can identify, target, and measure reliably.
• Re-segment when behavior patterns or business needs change materially.

Common Mistakes

• Building segments that are interesting but not actionable ("users who value sustainability").
• Running tests only on aggregate results, missing Simpson's Paradox where overall flat hides segment-level effects.
• Treating segments as static when behavior migrates across them.

Industry Context

SaaS and B2B segment by plan tier, usage depth, and account size (SMB, mid-market, enterprise). Ecommerce and DTC segment by RFM (recency, frequency, monetary), category affinity, and discount sensitivity. Lead gen operations segment by lead source, firmographic ICP fit, and sales-cycle stage.

The Behavioral Science Connection

Segmentation recognizes that customers can differ in context, needs, and behavior. Aggregate experiment results can obscure heterogeneous effects, but a behavioral segment is not automatically more predictive than a demographic one. Compare segment definitions on out-of-sample prediction, stability, actionability, and incremental decision value.

Key Takeaway

A segment is only useful if you can identify, target, and measure it — everything else is a persona deck that gathers dust.