Can segmentation be discriminatory?

Yes. Segmentation can produce discriminatory outcomes even when the intent is completely neutral. It's one of the less obvious risks in marketing data work, and it's worth understanding before you build complex targeting logic.

Direct discrimination is the obvious version: explicitly excluding or targeting people based on protected characteristics like race, gender, religion, age, or disability. This is illegal in regulated industries and generally a bad idea everywhere.

Proxy discrimination is the harder one to spot. It happens when seemingly neutral criteria (ZIP code, income bracket, education level) correlate strongly with protected characteristics. If you're targeting subscribers in affluent ZIP codes for a financial product, you might be inadvertently excluding racial or ethnic groups at a disproportionate rate, even though you never asked about race. The effect is discriminatory even if the intent wasn't.

Industries with regulatory oversight face the most scrutiny here. Fair lending, fair housing, and employment advertising laws specifically prohibit targeting decisions that produce disparate impact on protected groups. If you're in financial services, housing, or employment advertising, talk to your legal team before building exclusion-based segments.

For everyone else, the practical check is: audit your segments periodically for whether certain groups are systematically excluded from high-value opportunities. If your "best offer" segment happens to exclude a demographic group at a much higher rate than your general list, that's a signal worth investigating.

The safer approach is building segments around behavior and preference rather than demographic assumptions. Someone who clicked your pricing page last week is a better signal of purchase intent than someone who lives in a certain ZIP code. Behavioral criteria tend to be more accurate and less likely to produce discriminatory effects.

For the specific case of sensitive demographic attributes like gender, location, and income in segmentation, the sensitive attributes guide goes into more detail. And if you're combining behavioral data in ways that could raise discrimination concerns, the broader privacy implications of segmentation article covers the regulatory context.