Research quality & methodology

Know what your
results can tell you.

Useful research starts with the right people and a clear question. It also means understanding how answers are collected, how AI helps, and where the evidence stops.

Start with your audience

Your participants. Their answers.

You can share a survey link with customers or prospects you already reach. That gives you evidence from the people who respond; it does not automatically represent your whole market.

Managed recruiting is not yet available as a connected self-service option. Contact us to confirm audience availability, supplier arrangements, and the full project quote before paying for research that needs outside participants.

Collection & checks

Build care into the study.

The checks available depend on the test and collection flow you use.

01

Relevant participants

Define who should answer. Screening questions, when enabled, can check eligibility before someone takes the study. Review the screeners before launch.

02

Clear participation

Participant notices and consent checks explain the study before responses are submitted. Avoid requesting personal or sensitive information you do not need. The general release is intended for adult participants.

03

Valid submissions

Available response flows check required answers and whether selections fit the configured questions. Submission checks help prevent the same submission being counted again when a request is retried.

04

Review before relying

Review usable response counts, incomplete answers, unusual patterns, and study limitations. Automated validation does not establish that each person is unique or that every answer is attentive.

Sample size & interpretation

There is no universal “enough.”

Plan around the decision

The appropriate sample depends on the method, differences you need to detect, audience, and number of groups you want to compare. A total of 200 responses may leave much smaller groups for individual comparisons.

Agree the target and usable response requirements before collecting data. Report the denominator for each finding, rather than treating every percentage as if it used the full sample.

Keep the limits visible

Convenience samples and customer lists can miss important parts of a market. Stated purchase interest is not observed buying behavior. Small differences do not establish a reliable winner on their own.

A revenue illustration needs explicit assumptions about the audience size and conversion from interest to purchases. It is not a sales forecast or a profit estimate.

The role of AI

AI helps you prepare and explain.

Drafting with your review

AI can help turn your business question into a research plan and draft screening questions or explanations. Review those drafts for accuracy, relevance, and appropriate wording before using them.

People supply the evidence

AI-generated text is not a substitute for participant responses. Findings should be assessed against the chosen method, actual usable answers, and their limitations. AI interpretation can be wrong and requires your review.

Research planning and screener drafting use OpenAI. Avoid placing personal identifiers or confidential information into prompts unless it is needed and appropriate. See our privacy policy for more detail.

See the evidence in context.

Our fictional sample shows the recommendation, response counts, assumptions, and limits together.

View the sample report