Why recruitment decides what research finds
The customer research hub puts it plainly: loyal customers who answer emails are rarely the people whose behaviour the business most needs to understand. A study of why trials fail needs people who abandoned a trial; a study of why customers leave needs people who have left. Every recruitment choice either keeps the study pointed at those people or drifts it towards whoever was convenient, and the drift is invisible in the findings unless someone wrote it down. When a finished round of interviews has produced only praise, the diagnosis for customer interviews that are not producing insight starts by comparing the participants with the screener for exactly this reason.
| Recruitment choice | What goes wrong when it is made carelessly | The step that handles it |
|---|---|---|
| Who the study is about | Participants described by demographics, so the people with the relevant experience are a minority | Describe the people by what they did and when |
| Where they come from | A list that reaches only engaged customers, presented as if it reached everyone | Choose the frame and note who it misses |
| How many of each kind | One group crowds out another, and no comparison is possible | Set quotas |
| How they are screened | People guess the qualifying answer, or the wrong people slip through | Write the screener; add the exclusions |
| What the invitation says | A message that reads as marketing, or one that primes the answers | Invite on a research footing |
| What happens on the day | Empty slots filled with whoever is available | Plan for no-shows and replacements |
| What is kept afterwards | Names and contact details mixed into notes and findings | Keep the recruitment log apart |
The method is the same whichever study it feeds. The customer interviews tutorial and the usability testing tutorial both rest on it, and a group discussion needs it more than either, because a single wrong participant shapes what everyone else in the room says. Where studies follow one another, the customer research programme playbook keeps a standing pool of people who agreed to research contact, so recruitment does not start from scratch each time. The tutorial on building a research participant pool builds that pool, keeping it apart from marketing lists and rotating invitations so the same members are not drawn for every study, while each study still screens and logs its recruits as this page describes.
Before you start
Have the research brief in front of you, written as the research design guide describes, with the decision, the questions and the groups the decision compares. Recruitment is where a vague brief shows: if nobody can say whose experience answers the question, stop and settle that first.
Find out which lists you may use. Customer records, CRM entries for lost prospects and past research participants are all Personal DataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier., and whether each may be used to invite people to research is a question for whoever owns data protection, not for the person running the study. Ask at the same time whether people who have opted out of marketing are to be excluded.
Agree the incentive, its form and who pays it. The hub's rule is that an incentive compensates people for their time in a form that does not bias the result, such as a voucher or a charity donation, rather than a discount on the product being researched, which attracts people who already like it.
Decide who will run recruitment and who will see the recruitment log. Ideally the people analysing the findings work from participant codes, not names. The recruitment logThe study's restricted record of who was approached, who qualified and took part, with screener answers, consent, sessions, incentives and a deletion date. is the one place where those codes are matched back to names.
The steps
Describe the people by what they did and when
Write one or two sentences that describe the participants by behaviour and recency, as the interviews tutorial recommends: "cancelled their subscription in the last few months", "requested a quote and bought elsewhere", "used the new booking flow for the first time". Demographics and personasA research-based profile of a typical customer type: goals, problems and buying behaviour. can describe who these people tend to be, but they are not the criterion. The jobs-to-be-done framework is a useful check: the people you want are the ones who were recently trying to get the job done that the decision concerns, whether or not they chose you. Recent behaviour matters because memory fades and people fill the gaps with what they think they should have done.
Choose the frame and note who it misses
A frame is the list or channel participants are drawn from. The common ones are a customer list segmented by behaviour, CRM records of lost deals, an invitation on the website or in the product, a research panel, and a community where the audience gathers. Each reaches some people and misses others: a customer list misses everyone who never bought, a site invitation hears only from people who visited recently and stayed long enough to see it, and a panel reaches people who sign up to take part in research. Choose the frame that reaches the behaviour in step one, and write a line in the brief naming who it misses. The audience research guide lists the data most businesses already hold, which is often where the frame for existing customers comes from.
Set quotas for each group the decision compares
If the decision compares groups, such as people who converted and people who did not, new and long-standing customers, or buyers and approvers in a business sale, recruit each group on purpose. Set a quota for each so that one group cannot fill the study, and decide in advance whether any other mix matters, such as region or device. Keep quotas few: every extra condition makes recruitment slower and the screener longer. The segmentation the business already uses is a sound place to look for the dividing lines, provided they are ones that change the answer to the research question.
Write the screener so the qualifying answer is not obvious
The screener is a short questionnaire that confirms each candidate fits. Its first rule is the hub's: keep it neutral so it does not reveal the answer that qualifies. Ask about behaviour with a full list of options rather than a yes-or-no question about the one you want; put the topic you care about among several others; ask when someone last did something, not whether they ever do it; and avoid naming your brand early, because people who know who is asking answer differently. Keep it short. Every question should either qualify, exclude or place someone in a quota, and anything else belongs in the session. The table below works through a screener question by question.
Add the questions that exclude the wrong people
Some people should not take part whatever their answers. The usual exclusions are people who work for the business or a competitor, people who work in market research or marketing in the category, and the friends and family of staff, because they know too much or have a stake in the answer; the usability testing tutorial makes the same rule for its sessions. Add a question that catches people who have taken part in research very recently, so the same voices do not recur. Place the exclusion questions early, so nobody spends time on a screener they cannot pass.
Write the invitation on a research footing
The invitation says who is asking and why, what taking part involves, how long it takes, whether it is recorded, what the incentive is and that taking part is voluntary. Those points come from the professional codes the site's research guides describe from the MRS and ESOMAR: participation is voluntary and informed, people are told who is collecting the data and why, and research is never a disguise for selling. The invitation therefore promotes nothing. As the marketing research hub notes, an invitation that promotes a product is marketing whatever its subject line says, and the comparison of promotional and transactional email shows how purpose, not the label, decides what a message is. Leave the topic general enough that it does not prime the answers, and send it only to the people the data protection owner has cleared; where they have asked for marketing opt-outs to be excluded, check the list against the suppression list first.
Screen, select and confirm
Read the completed screeners against the profile, the quotas and the exclusions, and select people who fit, favouring variety within each quota over the first to reply. Invite more people than you need, as the interviews tutorial advises, and keep the extra screened candidates as reserves rather than booking them all. Confirm each session in writing with the time, the format, what to bring or install, the recording and the incentive, and send a reminder before the session. Leave gaps between sessions for notes. At the start of each session, confirm the one or two facts that qualified the person; if they turn out not to fit, thank them, pay the incentive and end the session.
Plan for no-shows and replacements
Some people will not turn up and some will cancel. Decide in advance how a missing participant is replaced: a reserve from the same quota, invited from the screened pool, never whoever happens to be free. If a quota cannot be filled, say so in the findings rather than quietly filling it with someone from another group. Note every no-show and replacement in the recruitment log; a pattern, such as one group repeatedly failing to appear, is itself information about how the study was received.
Keep the recruitment log apart from the findings
The recruitment log holds names, contact details, screener answers, quotas, consent, sessions, incentives paid and replacements. It is personal data under the GDPR, whose principles, as the GDPR glossary entry summarises them, include collecting only what the purpose needs and keeping identifiable data no longer than necessary. Give each participant a code, use the code in notes and findings, keep the log where only the people running recruitment can see it, and agree its deletion date with whoever owns data protection. Record the date each person took part so the next study can apply a contact rule; the voice of the customer guide uses the same idea to stop customers being surveyed too often, regardless of which team wants to ask.
Once recruits are confirmed, the research session readiness checklist carries these recruitment lines into a gate with consent, materials, the pilot and notes, ticked before the first session.
A screener, question by question
The table shows the kind of question a screener contains and what each does. The wording is an example for a study of lapsed subscribers; adapt it to the profile in step one.
| Question | What it does | How it stays neutral |
|---|---|---|
| "Which of these, if any, do you or anyone in your household work in?" (a list including the category, market research and marketing) | Excludes people with a stake or inside knowledge | The excluded answers sit among unrelated industries |
| "Which of these services have you used in the past year?" (a list of several services) | Confirms category experience | The service being studied is one option among many |
| "When did you last use it?" (time bands) | Applies the recency criterion | Asks when, not whether; no band is marked as wanted |
| "Which best describes how you use it now?" (still using, paused, stopped, switched to another) | Places people in a quota or screens them out | All states are offered; none is singled out |
| "Have you taken part in a research interview or discussion recently?" | Stops the same people recurring | Asked factually, with no hint that one answer disqualifies |
| "Are you able to join a video call with your camera on?" | Confirms the format works for them | A practical question, asked last |
A screener that mixes these jobs well is short enough to finish without effort and tells the candidate nothing about which answers matter. Recruitment requirements belong in the audience and sample field of the research brief template, so the screener can be checked against them before it goes out. Where the study is a survey of a whole customer list rather than a set of sessions, sampling rules replace screening, and the survey design guide covers frames, samples and non-response. The methods hub sets out the selection and non-response biasesThe gap between the people a survey or study invited and those who answered; engaged, satisfied customers tend to reply more, so results lean their way. that recruitment is trying to limit.
A bicycle servicing subscription recruits lapsed members (illustrative)
Vardunel Cycles recruits lapsed members and people who never joined
Vardunel Cycles, an invented bicycle servicing subscription, wants to know why members lapse after their first year and why people who price the plan do not join. The brief compares two groups, so recruitment sets a quota of 6 lapsed members and 6 people who priced the plan in the last 3 months but never subscribed.
The first draft of the screener opened with "Did you recently cancel your Vardunel subscription?", which told every candidate what the study wanted and invited people who had cancelled years ago to say yes. The rewrite asks which of 8 bike services people have used in the past year, when they last used each, and how they use them now, and only names the brand in the session. Two exclusion questions screen out 4 candidates who work in bike shops and 1 whose partner works for the business.
The data protection owner clears lapsed members whose records allow research contact and excludes anyone who opted out of marketing, which removes about a fifth of the lapsed list. People who priced the plan are reached through a short invitation on the pricing page that says the study is research and promotes nothing. 41 people complete the screener; 12 are booked and 5 kept as reserves. Two lapsed members do not join their sessions, and both slots are filled from the lapsed reserves rather than from the pool of people who never joined. The recruitment log records the no-shows, the replacements and a deletion date, and the notes use codes from L1 to L6 and P1 to P6.
Verify it worked
- Compare each confirmed participant's screener answers with the profile and the quotas. Everyone should fit one group, and each group should be filled, as with the 6 lapsed members and 6 people who priced the plan.
- Show the screener to a colleague outside the study. They should not be able to tell which answers qualify.
- Read the invitation. It should promote nothing and go only to people the data protection owner has cleared.
- Open the recruitment log. It should hold the names, codes and deletion date, and the notes should carry codes only.
- Watch the start of each session. The one or two qualifying facts are confirmed before the questions begin.
If it goes wrong
- Nearly every candidate gives the qualifying answer. The screener gives it away. Rewrite it with a full list of options, the topic among others, and "when" rather than "whether".
- The interviews produce only praise. The frame reaches engaged customers. Compare the participants with the screener, then choose a frame that reaches the behaviour in step one.
- A quota cannot be filled. The frame may miss that group. Say so in the findings and do not fill it with someone from another group.
- Someone does not turn up. Invite a reserve from the same quota, never whoever is free, and note the no-show in the log.
- A participant at the session does not fit the profile. Thank them, pay the incentive and end the session. Then check which screener question let them through.
Next steps
- Move on to the customer interviews tutorial when the confirmed sessions are interviews.
- Use the usability testing tutorial when the confirmed sessions are tests of a flow or page.
- Read the survey and questionnaire design guide if the question needs a whole customer list rather than a set of sessions.
Common mistakes
- Recruiting whoever answers. The engaged customers who reply first are the people the question is least likely to be about.
- Describing people by demographics. An age band and a region do not find the people who recently did the thing the decision concerns.
- A screener that gives the answer away. A yes-or-no question about the exact behaviour you want tells candidates how to qualify.
- No exclusions. Staff, their families and people who work in the category or in research take part and answer like insiders.
- An invitation that sells. A research invitation with an offer in it is marketing, and it recruits people who came for the offer.
- Filling empty slots with anyone. A replacement from the wrong group quietly changes the comparison the study was built to make.
- Names in the findings. Contact details and screener answers belong in a restricted log with a deletion date, not in notes the whole team reads.
