Choose when a segmented read beats an overall average.
Averages are useful when the population is meaningfully homogeneous. They mislead when different segments respond in opposite directions. The mechanism is aggregation masking: the total effect can hide where the decision should differ by customer, channel, geography, or starting behavior. Data-driven decision-makers do not segment endlessly after the fact, because that creates false discoveries. They pre-name the segments that have a plausible mechanism, then use them to decide whether to target, iterate, or stop. The question is not “Can I find a subgroup that looks good?” It is “Did we expect this subgroup to behave differently, and is the effect…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in