How Do I Recruit the Right Users for Testing My Mobile App?

Recruitment is the part of user research that teams treat as logistics and then wonder why their findings feel hollow. You book five people, run the sessions, fill the report, and ship a product that still confuses the people it was built for. The problem rarely sits in the sessions themselves. It sits in who walked through the door before a single question was asked.

Good recruitment is about finding people whose behaviour will actually answer your research questions.

Getting recruitment right is harder than it looks because the obvious answer, find people who match your target audience, is true but incomplete. The right participants are the people whose behaviour, attitudes, and mental models will actually answer your research questions. Those two things overlap a lot, but not always, and the gap between them is where research quietly fails.

We have seen this play out across products in travel, property, logistics, and film. The pattern is consistent: teams that think carefully about who they are recruiting, and why, come out of research with things they can act on. Teams that treat recruitment as a box to tick come out with data that confirms what they already believed. This article is about how to do the former.

Why the Wrong Users Ruin Good Research

Imagine running a session with someone who has never booked a holiday online, when you are testing a travel booking app aimed at frequent independent travellers. They will struggle with flows that your actual users find natural, flag problems that do not exist for the real audience, and miss the friction points that genuinely matter. Your findings will be distorted, and you will not know it until you ship.

The damage runs in both directions. Participants who are too expert can be just as misleading as participants who are too inexperienced. Power users skip steps that new users cannot, forgive errors that real users would abandon over, and articulate preferences that reflect their habits rather than the product's design. Research built on the wrong participants does not just produce bad data. It produces confidently wrong data, which is harder to recover from than no data at all.

According to Springer's BMC Medical Research Methodology, 81% of focus group studies reviewed acknowledged limitations relating to recruitment, with biased participant selection cited as the most common problem. That figure comes from health research, but the dynamic it describes is familiar in product and UX work. Bias in who you recruit shapes everything downstream, and teams typically acknowledge that risk only after the sessions are done.

Define Who You Are Actually Testing For

Before you write a screener or contact a recruitment agency, you need to answer a simpler question: what are you testing, and what kind of person's experience will actually tell you what you need to know? These are different questions to "who is our target audience", and conflating them is where recruitment goes wrong early.

A product can have a clearly defined target audience and still need research participants who are different from that audience at different stages. Early-stage concept testing benefits from people who have the problem you are solving, regardless of whether they fit your persona exactly. Usability testing on a built product needs people who closely match your actual users. Emotional and attitudinal research sometimes needs people who embody qualities you are trying to design into the product, even if they would never become customers.

The practical starting point is to write a one-sentence research question for each session type, and then ask honestly: what background, behaviour, and context does a participant need to have to give us a meaningful answer? That question drives the criteria. Demographics come later, and they matter less than researchers typically assume.

When the Right Users Are Not Your Target Users

One of the most useful things we have done in recruitment is bring in participants who would never become customers, because those participants had something we could not find in the target audience.

On a concierge app we built for a premium high-rise property development, the target customers were high-net-worth residents, people who value privacy, tend not to interact with neighbours, and have little experience of the kind of community feeling that characterises tightly-knit residential areas. We ran focus groups with that group, but we also brought in people from social housing, who were never going to be customers. We did it because they had exactly what we were trying to build into the product: a genuine, natural sense of community, neighbourliness, and shared belonging.

Non-target participants can carry qualities you want to design in that your actual audience cannot articulate.

What those sessions revealed was that everything came down to communication, residents in the target group simply did not know their neighbours and had no channel to change that. The social housing group showed us what that sense of connection actually looked like in practice. We redesigned the product around it: automating the mundane tasks so the concierge and residents had more time for genuine interaction, and adding personalisation features so residents could see who their neighbours were and what they cared about.

The point is not to recruit randomly. The point is to ask what quality or behaviour you need to observe, and then find the people who have it, whether or not they are your target user.

Writing a Screener That Filters Accurately

A screener is the questionnaire you send to candidates before inviting them into a session. Its job is to include the right people and exclude the wrong ones, and it fails at that job more often than teams realise, usually because it is written to describe a persona rather than to identify a behaviour.

Ask about behaviour, not self-description

The most common screener mistake is asking people to rate themselves. "How comfortable are you with technology?" produces inflated answers. "How many apps did you download in the last three months?" produces observable behaviour. Questions grounded in frequency, recency, and specific actions filter far more accurately than questions about attitude or confidence.

Build in deliberate disqualifiers

A screener should not only find the right people. It should actively remove the wrong ones. Include questions where certain answers should disqualify a candidate, people who work in UX or market research, people who have tested apps professionally, people who use a competitor product exclusively. Be explicit about what disqualifies rather than leaving it to interpretation.

Write your screener in a random order so that the "right" answers are not obvious. If candidates can see what you are looking for, they will shape their responses accordingly, and you end up with participants who passed the screener rather than participants who fit the criteria.

Keep screeners short. Twelve to fifteen questions is usually enough. Beyond that, genuine candidates drop out and only the highly motivated, who are often atypical, complete it.

Behavioural and Attitudinal Criteria That Matter More Than Demographics

Age, location, and income bracket are the first things that go into a recruitment brief and often the least useful. They describe who someone is, not how they behave or what they believe, and behaviour and belief are what drive the interactions you are trying to study.

On a travel product we worked on, the finding that changed the design had nothing to do with user demographics. It was about mental models, specifically, how users expected to see pricing presented. Users expected a platform fee to appear as a separate line item, because that matches what they had seen across every other travel and booking app they had used. We had initially wrapped the Stripe fee into the total price, thinking a single clean number would reduce friction. Instead, it created anxiety: users assumed a hidden charge would appear at checkout. The moment we broke out the fees transparently, trust increased, even though users were looking at more information.

That insight came from recruiting people based on their booking behaviour and familiarity with travel platforms, not their age or income. The relevant attitudinal criterion was their existing mental model of how platforms present costs. Demographic filters alone would never have surfaced it.

When writing recruitment criteria, list the three or four things a participant must have experienced or believed to give you a meaningful answer. If those things are not on your screener, add them before you send it out.

How Many Participants Do You Actually Need

The number question comes up in almost every project conversation, and the honest answer is: fewer than you think, as long as they are the right people.

Nielsen Norman Group has long argued that testing with five users uncovers around 85% of usability problems, because each additional participant beyond that point largely surfaces the same issues you have already seen. That principle applies to qualitative usability testing specifically, but it points to something true across most research formats: the return on adding more participants drops sharply after a relatively small number, and the energy is better spent improving who you recruit than how many.

Adjust for audience complexity

Five participants works when your audience is reasonably homogeneous. If your product serves two or three genuinely distinct user groups, say, the driver and the customer on a delivery platform, or the resident and the concierge on a property app, you need five per group, not five total. The number scales with the number of meaningfully different audiences, not with the overall size of your user base.

Plan for drop-outs

Recruit one or two more participants than you need for each session block. People cancel, misread times, or fail a final eligibility check on the day. If you have planned for five and three show up, you do not have a study. If you have planned for seven and five show up, you do.

Recruiting Methods and Where to Find the Right People

The method you use to find participants shapes who you end up with. Different channels attract different types of people, and the mismatch between channel and audience is a common source of recruitment problems.

According to Springer's BMC Medical Research Methodology, 48% of focus group studies used direct approach as their primary recruitment method, reaching out personally to people identified as fitting the criteria. That works well when you have access to a relevant community or existing customer base. For product teams, it often means recruiting from your own user list, which is fast and produces well-matched participants, but risks over-representing your most engaged and forgiving users.

The main channels worth knowing are laid out below.

No single method is always right. The best approach is usually to combine two: a primary channel that gives you volume, and a secondary channel that fills the gaps your primary misses.

Incentives, Honesty and the Problem of Professional Testers

Incentives increase participation rates, and that is a good thing when you are struggling to fill sessions. The issue arises when incentives attract people whose primary motivation is the reward rather than the research, and who have learned to say the right things to pass screeners.

Professional testers are a real problem on recruitment panels. These are people who participate in research sessions regularly, sometimes daily, and who have developed an intuitive sense of what researchers are looking for. Their responses can sound thoughtful and usable while reflecting performance rather than genuine reaction. The higher the incentive, the more they are drawn in.

According to Springer's BMC Medical Research Methodology, 43% of focus group studies reviewed used incentives to improve participation. That is a reasonable proportion, and incentives are worth using, the risk is in how they are structured, not whether they exist.

Ask screener candidates how many research sessions they have participated in over the last six months. Anyone who has done more than three or four is worth treating carefully. It does not automatically disqualify them, but it is a signal worth acting on.

Non-cash incentives, gift cards, product access, charitable donations in the participant's name, tend to attract people with a genuine interest in the subject over people who are purely chasing cash. The right incentive level is enough to cover people's time without making participation feel like paid work.

Validating Your Recruitment Before Sessions Begin

Recruiting participants and confirming they are the right participants are two different things, and teams often treat them as one. By the time a session starts, it is too late to discover the person in front of you does not match your criteria. Validation happens before that.

Run a brief pre-session check

A short five-minute call or message exchange before the session date gives you a chance to confirm the key criteria verbally. You are listening for consistency, does what they tell you now match what they told you on the screener? Do they reference the behaviours and contexts they described? If something feels off, it usually is, and replacing a participant before the session is far less disruptive than discovering the mismatch halfway through it.

Use an observation protocol during sessions

Brief whoever is observing your sessions, whether a colleague, a stakeholder, or a note-taker, to flag participants who seem significantly out of step with the target criteria. Sometimes a participant passes the screener honestly but turns out to have a context or background that makes their responses hard to interpret. Noting that in real time means you can weight it appropriately when you analyse findings, rather than letting it quietly skew the results.

On the film industry document portal we worked on, we were designing for a diverse group of people operating under serious time pressure on set. Validating that participants actually worked in fast-paced production environments, not just adjacent to them, was the difference between research that reflected real conditions and research that reflected an idea of those conditions.

Conclusion

Recruitment is where research is won or lost, and the decisions that matter most happen before sessions start. Who you invite in, how you filter them, and what criteria you actually use to select them will shape everything you learn, and everything you subsequently build.

The most common errors are all variations on the same mistake: treating recruitment as a logistics task rather than a design decision. Defaulting to demographics over behaviour. Assuming target audience and right participant are the same thing. Skipping validation because the sessions are already booked. Writing screeners that signal the answers you want.

The concierge project we worked on showed us clearly that the right participants sometimes come from outside your target audience entirely. The travel product showed us that mental models matter more than demographic profiles. The logistics platform showed us that recruiting separately for each user group, driver and customer, produces completely different findings to treating them as one.

The investment in getting recruitment right is modest compared to the cost of building from bad research. Five well-recruited participants will give you more than fifteen poorly recruited ones, and the quality of your screener, your criteria, and your validation process is what makes that true.

If you are about to kick off user research and want to make sure you are bringing in the right people, let's talk about your recruitment approach.

Frequently Asked Questions