Superhuman: Find the people who need you most.

Imagine you manage a new reporting tool. Some people rely on it every afternoon, while others tried it once and left. A single satisfaction average gives the team an easy number but a difficult decision. Should you broaden the feature set or deepen the experience for the people who already depend on it? Start by defining what you need to learn, rather than choosing the answer with the largest audience.

Our fictional learner, Leena, coordinates five client updates each day. A missed message can block another person’s work. Compare her situation with someone who prepares one personal report a month. These people may want similar-looking screens for very different reasons. A segment becomes useful when it explains a recurring need, constraint, or benefit. Calling someone a power user only restates the activity you observed.

Use the model below to inspect three fictional survey groups. Notice that a higher support percentage can come from a smaller selected group. That can be a promising lead, but selecting the best-performing group after seeing the data creates a hypothesis to validate. It does not turn a small survey into proof of a broad market. Keep eligibility, response counts and nonresponse beside every percentage.

Suppose daily coordinators consistently value reliable handoffs. You might prioritise that workflow, explain it more clearly in onboarding, and recruit more people in the same situation. That choice also postpones other needs. Write down who benefits, who waits, and what would justify expanding later. Avoid assuming that every complaint from a less-engaged group is irrelevant; accessibility and safety problems can cross segment boundaries.

For your project, combine the survey lead with an observed task and a meaningful return event. Ask which benefit people would miss and what alternative they would use. Check whether the promising group remains promising in a fresh sample. Keep acquisition channel and tenure visible. If the story depends on excluding inconvenient responses without a principled eligibility rule, revisit the claim before presenting it to the team.

Create a one-page segment brief: situation, repeated need, current workaround, main benefit, source evidence, and missing evidence. Add one comparison group and a fresh-sample validation plan. Keep survey counts separate from forecasts. Your deliverable is a focused research decision, not a declaration that you have found product-market fit.