How to Recruit Users for Testing: A Practical Guide
Define representative users, write a screener that cannot be gamed, choose a recruiting source, and buffer for the 11% no-show rate. A practical guide.
By GhostClick / Method / Published 15 Jul 2026 / 7 min read
Learning how to recruit users for testing is the part of research that quietly decides whether the rest of it works. The method can be flawless, but if you talk to the wrong people, or too few of them, the findings are noise dressed as insight. It is also the step teams most often underestimate: in Jakob Nielsen’s survey work for the Nielsen Norman Group, in-house recruiting took an average of 1.15 hours of staff time per participant, and the average no-show rate ran at 11 percent, roughly one in nine. Recruiting is real work, and it rewards a real process.
This guide walks that process in order: define who counts as a valid participant, screen for them honestly, choose where to find them, plan for the people who will not turn up, and stay on the right side of the ethics. It is vendor-neutral. Where a tool genuinely helps, this guide says so, and it also says where one does not.
Start by defining the user, not the headcount
The first question is not “how many?” but “who?” Nielsen’s first rule of recruiting is to get representative users: people whose behaviour and context match the ones your product actually serves. Recruit the wrong users and the study cannot be salvaged by sample size or clever analysis.
Write down the specific behaviours and circumstances that make someone a valid participant for this study. Not demographics for their own sake, but the things that change how a person would use what you are testing: whether they have done the task before, which tools they use now, the constraint they are under. This definition becomes your screener.
A participant is representative when their real behaviour, not their job title, matches the situation you are designing for.
Screen behaviourally, and hide the answer
A screener is a short set of questions that decides who qualifies. The Nielsen Norman Group is blunt that using a screener is the most reliable way to end up with the right participants. Two rules make a screener work.
First, do not telegraph the answer you want. If a question makes the study’s purpose obvious, some people will shape their answers to qualify. Avoid leading questions that imply the desired response. Second, ask about behaviour, not self-assessment. Asking “how often do you play online games?” invites exaggeration; asking someone to name their favourite games and say why quickly separates the committed player from the one who barely remembers any. Behavioural questions are harder to game because they ask for something a real user would know and a pretender would not.
- Lead with a soft disqualifier so unqualified people drop out early, not after they have invested time.
- Mix in a plausible wrong option or two, so “selecting all the right answers” is not the obvious path to acceptance.
- Keep it short. A screener is a filter, not the interview. Five to eight questions is usually enough.
Choose where to find them
There is no single best source. There is the source that best fits your audience, budget, and timeline. Most teams recruit their own participants: in Nielsen’s survey only 36 percent used an outside recruiting agency and just 9 percent relied on one exclusively. The trade-offs are worth naming plainly.
| Source | Best when | Watch out for |
|---|---|---|
| Your own users | You need people who already use the product or a close competitor. | Selection bias toward your most engaged, forgiving users. |
| Recruiting agency | You need a hard-to-reach audience fast and have the budget. | Professional participants who test for a living; screen harder. |
| Research panel | You run studies often and want a standing pool to draw from. | Panel fatigue and drift from your real audience over time. |
| Intercept / in-context | You want people in the real moment of use, on-site or in-app. | Self-selection: only the willing and unhurried respond. |
Whichever you choose, the screener does the quality work. The source determines reach and cost; the screener determines whether the people you reach are the right ones.
Recruit more than you need
Plan for the no-shows before they happen. With an 11 percent average no-show rate, a study that needs five participants will, often enough, lose one. The standard fix is to over-recruit: the Nielsen Norman Group calls the extras floaters, backup participants booked so that every research slot gets filled even when someone cancels or their data has to be thrown out.
Over-recruiting also covers the participant you discover mid-study was not representative after all, or who clearly gamed the screener. In a small qualitative study, every lost session is a large share of your total insight, so the buffer is cheap insurance. How many you need in the first place is its own question, covered in our guide to how many participants qualitative research needs. In short: around five for a qualitative usability pass, and 20 to 40 when you need statistics.
Treat incentives and consent as part of recruiting
Recruiting does not end at “yes”. Offer an incentive that respects people’s time and is fair across the sample, so you are not quietly selecting for whoever is cheapest to reach. Be clear about what the session involves before anyone commits, and get informed consent, including consent to record, in writing. A participant who knows what they signed up for is a participant who shows up and engages.
This is also where a well-written invitation earns its keep. People skip sessions they do not understand or trust. A specific, honest invitation that names the time, the format, and the incentive reduces the very no-show rate you are budgeting for.
Where a tool fits, and where it does not
Recruiting is sourcing plus screening plus getting people into the session. GhostClick does not do the first two parts: it has no participant panel and no screener, so finding and qualifying the right people is still your job, exactly as it is with a human-run study. What GhostClick does handle is the last part. Once you have your list, it gives each participant a unique private link, or emails the invitations for you in one go, then runs the interview itself by text or voice in seven languages, on the participant’s own schedule. That removes the scheduling and moderating load, not the recruiting load.
And it is the wrong tool for some jobs. If your study needs a moderator reading body language and improvising in the room, or you are doing generative, deeply exploratory work where the whole point is to follow the unexpected in real time, an asynchronous AI-moderated interview is not the right fit. It is strongest for structured discovery and evaluative sessions at a scale that live interviews cannot reach. The honest comparison lives in our piece on async AI-moderated versus live interviews, and the questions themselves matter as much as the people, which is the subject of writing a discussion guide that gets honest answers.
Bring your participants, skip the scheduling
GhostClick gives each person you recruit a private link, or emails the invites for you, then interviews them by text or voice in seven languages on their own time. You recruit; it runs the sessions.
Start a studyFrequently asked questions
- How do you recruit participants for user testing?
- Define the behaviours that make someone representative, write a short screener that hides the study purpose and asks about behaviour rather than self-assessment, choose a source that fits your audience and budget, and over-recruit to cover no-shows. The screener does the quality work; the source only determines reach and cost.
- How many participants should you recruit for usability testing?
- For a qualitative usability study, about five participants is the standard working number, while quantitative studies need roughly 20 to 40. Because the average no-show rate is around 11 percent, recruit about 20 percent more than your target so every slot is filled.
- What makes a good screening question?
- A good screening question asks about real behaviour instead of self-assessment and does not reveal which answer qualifies. Asking someone to describe their favourite tools and why separates genuine users from pretenders far better than asking how often they do something, which invites exaggeration.
- Should you recruit your own users or use an agency?
- Most teams recruit their own participants: in Nielsen Norman Group survey work only 36 percent used an outside agency and just 9 percent relied on one exclusively. Use your own users for people who already know the product, and an agency when you need a hard-to-reach audience quickly and have the budget to screen out professional testers.
Related guides