Start with the decision, not the questionnaire
A research project often begins with a broad question such as “What do customers think of this service?” That question may generate interesting responses, but it gives little guidance on what to ask or how to interpret the answers. Before collecting data, name the decision the findings will inform. For example: “Should we simplify the sign-up process before promoting the service to new customers?”
Next, list the realistic actions available to the team. It might keep the current process, revise one confusing step, or redesign the entire flow. Success criteria should help distinguish among those actions. They are not a promise that the research will produce a convenient answer; they are an agreement about what kind of evidence would make each action reasonable.
Describe what success would look like
Separate success for the research project from success for the product or service. A successful study may reveal that an idea needs substantial work. For the study itself, useful criteria might include reaching the intended audience, learning where people encounter difficulty, and producing findings specific enough to guide a change. For the product, a team might care about whether people can complete a task or understand an offer. Keeping these definitions separate prevents disappointing findings from being dismissed as a failed study.
Make each criterion observable. “People like the sign-up process” is hard to assess because liking can mean several things. “Participants can explain the next step without prompting” is clearer. If a survey is appropriate, the team could also define which question will measure perceived clarity and which responses would signal concern. The measure should match the decision: a rating alone may not show which step needs fixing, while an interview alone may not indicate how widespread a concern is.
Set evidence standards before seeing results
Decide in advance whose experience matters for this decision. Current customers, recent drop-offs, and people encountering the service for the first time may describe very different problems. If the proposed action concerns new customers, a study dominated by experienced users could miss the relevant friction. Specify any essential audience groups and note whose views the study will not represent.
Then define the strength of evidence needed for the decision. For an exploratory interview study, success might mean identifying recurring points of confusion and understanding the circumstances behind them, not estimating their prevalence. For a survey, the team might agree on a response pattern that warrants a closer look, while considering whether the respondents adequately reflect the audience of interest. Avoid treating a small difference or a vivid comment as decisive simply because it confirms an expectation. Record what would count as inconclusive, too, and what additional work would be needed in that case.
Turn criteria into a simple decision guide
A short decision guide can connect findings to actions without reducing people’s experiences to a single score. For the sign-up example, the team might agree to keep the process if intended users can complete it and explain what happens next, to revise a particular step if confusion clusters there, and to investigate further if different audience groups struggle in different ways. These are conditional plans, not conclusions written before the study.
Include practical constraints in the guide. A suggested change may be valuable but difficult to deliver immediately; another may be small enough to test quickly. State who will review the evidence, when a decision will be made, and what information they need to see. This keeps success criteria connected to an actual choice rather than a report that has no clear owner.
Leave room to learn something unexpected
Predefined criteria protect against moving the goalposts after results arrive, but they should not prevent the team from noticing a problem it failed to anticipate. If interviews reveal that people understand sign-up perfectly but distrust a later message, document that finding separately. Explain whether it changes the original decision or calls for a new question. Do not quietly replace the original criteria to make the project appear more successful.
Before launching the study, write the decision, audience, measures, evidence standards, possible actions, and known limitations in a brief plan. Ask colleagues involved in the decision to review it while changes are still easy to make. When the findings arrive, compare them with that plan, note any surprises, and explain the reasoning behind the next step. That process makes research easier to use even when the answer is mixed.
