Participant

Origin & evolution

The discipline of choosing how many participants to recruit, and which ones, sharpened around a single famous claim. In 2000 Jakob Nielsen published Why You Only Need to Test with 5 Users, drawing on a mathematical model he had developed with Thomas Landauer in 1993. The argument was that five users in a qualitative usability test surface roughly 85% of the problems, because problem discovery follows a diminishing-returns curve.

The claim was widely misread as "five is always enough", and Nielsen spent the rest of the same article qualifying it. The five-user figure holds for a homogeneous user group running a single qualitative test; a site serving several distinct audiences needs🫀NeedUserA user need, pain, desire, or constraintView reference → separate small samples per group, and quantitative measurement needs far more (NN/g now recommends around 40 for statistics). The number is a budget heuristic, not a law.

The wider qualitative-research field reached a parallel answer through saturation: keep recruiting until new participants stop adding new themes. A 2021 systematic review by Hennink and Kaiser found saturation typically arrives between nine and seventeen interviews for narrowly defined, homogeneous studies, and that diverse populations need roughly 56% more participants on average to get there. Sample size, in both traditions, is a function of how varied your population is and how broad your question.

How it works in practice

A team planning a discovery study on invoicing wants people who have abandoned an invoice💳InvoiceSales & RevenueAn invoice for billingView reference → in the last month. They write a screener: three closed questions that filter for the behaviour, plus one open question to catch professional testers who game incentives. They recruit twelve participants, slightly above the saturation floor, because the population spans freelancers and finance staff.

Each participant signs a consent form covering recording and data use, and receives a 60-pound incentive calibrated to the 45-minute session and the seniority of the people they want. By the eleventh interview no new abandonment reason has appeared for two sessions running. The team stops, confident the sample has earned its conclusions because recruitment screened for the behaviour under study rather than for whoever answered the panel first.

Participant vs. its neighbours

• Persona🎭PersonaUserAn archetype representing a user segmentView reference → is a synthesised, fictional composite built from many participants. A participant is one real, recorded individual. You recruit participants and you author personas; conflating them lets a single loud interviewee masquerade as a whole segment.
• Quote is the verbatim evidence🔎EvidenceValidationData supporting or refuting a hypothesisView reference → a participant produces during a session. The participant is the source; the quote is the artefact. One participant voices many quotes, and a quote is only as trustworthy as the participant behind it.
• Research study is the container that enrols participants against a plan. The study defines who qualifies; the participant is an instance that meets the bar. A study with weak screening criteria recruits the wrong participants no matter how many sign up.

In the graph

In the Unified Product Graph, a participant sits in the user-research region as the recruited subject of a study. A research_study_enrolls_participantResearch StudyenrollsParticipanthierarchy edge ties each person to the study they joined, research_plan_recruits_participantResearch PlanrecruitsParticipantcausal records the screening intent that admitted them, and participant_voiced_quoteParticipantvoicedQuotecross-domain links them to the verbatim evidence they produced. That structure keeps provenance intact: every quote and every insight🔦InsightUser ResearchA synthesised finding from researchView reference → traces back to a named participant and the study that recruited them, so a team can always ask who said this, and under what study design.