How to recruit participants for AI-moderated customer interviews
By Hey Juno · Published · Updated
To recruit participants for AI-moderated customer interviews, choose the route that matches who you need to hear from. Invite your own customers through a link, use a research panel for people you cannot reach yourself, or connect a recruitment partner you already work with. Then screen carefully, interview enough people to see patterns, and check quality before you analyse.
This guide covers participant recruitment for qualitative research run by Juno, the AI researcher from Hey Juno.
Start with whose story you need to hear
Good research starts before the first question, with who you invite. A study can only report the experience of the people who took part, so recruitment decides whose voice the findings carry.
Write down the people whose answers would change your decision. "Customers" is rarely precise enough. "People who cancelled in the last 90 days" or "decision makers who compared at least two financial platforms" gives you something you can screen for.
If you hold their contact details, ask them directly. If you can describe them but cannot reach them, you need a panel or a recruitment partner.
The three main routes
Your own customers, through invite links
Every Juno study has its own invite link. You send it by email, in a message, in a community or wherever you already talk to people. Anyone who opens it starts the interview in their browser, with nothing to download. You can create per-person links in Juno that carry context from your customer records (CRM) and notes for the interviewer.
This is usually the cheapest route, and people who know your brand tend to have things to say. The trade-offs:
-
You only reach people who already deal with you. Competitors' customers and non-customers are out of reach.
-
Willing responders differ from the quiet middle of your base.
-
You send the invitations. Juno does not do it for you.
A research panel
A research panel is a pool of people who have agreed to take part in studies for a reward. Teams use one to reach strangers who match a profile.
Hey Juno offers a panel service reaching more than 21 million people in 56 countries through verified panel partners. It is a managed service: a specialist scopes the study and recruits people who match your criteria. Going live does not start recruitment on its own.
The trade-off is cost and control. Recruitment and incentives are priced separately from interviews, and a panel is only as good as its screening and quality checks.
Your own recruitment partner or panel provider
If you already work with a recruitment firm or a panel provider, you can connect it to your study. You configure a participant ID and a link to send people back to your provider when they finish, so the provider knows who completed.
This suits agencies and larger teams with established suppliers. The cost is setup: you configure the identifier and completion destinations before sharing the provider's link.
Choosing between them
| Route | Best when | Watch for |
|---|---|---|
| Invite links | You hold the contact details | Sample limited to people who already know you |
| Research panel | You need strangers who fit a profile | Cost, and the need for tight screening |
| Your own partner | You have an existing supplier | Setup, and checking their quality controls |
Studies often mix routes. As an illustration, a churn study might invite recent leavers directly and add a panel group of people who left a competitor, so the two can be compared.
Screen for what you need to know
Screener questions decide who enters the interview. In Juno they are called pre-interview questions. Good ones are factual and hard to guess. Ask what someone did, how recently and how often, not whether they are "interested in" a topic.
A few habits help:
-
Ask about behaviour. "Which of these have you used in the past month?" beats "Do you care about energy bills?"
-
Include options that do not apply, so nobody qualifies by agreeing to everything.
-
Keep it short. Each question is effort for the participant.
-
Do not make the target obvious, or people will pick it.
Rehearse before launch to catch weak wording or screening problems. See real participants versus synthetic respondents for how rehearsal differs from research.
How many people to interview
There is no single correct number, and any figure should come with its conditions. Research on thematic saturation, the point where new interviews stop adding new themes, gives a rough guide. Guest, Bunce and Johnson ('How Many Interviews Are Enough?', Field Methods, 2006) found that most themes appeared within about twelve interviews in a fairly uniform group. Hennink and Kaiser (Social Science & Medicine, 2022) reviewed empirical studies, mainly of uniform samples with narrow aims, and found saturation reached within about 9 to 17 interviews. Understanding themes fully often takes more.
Those figures describe small, uniform samples. Treat them as a floor. Plan for more if you want to compare segments, the audience is varied or the decision is expensive to get wrong.
Hey Juno's priced-per-interview model means teams often run well beyond the saturation figures. Its published case studies include studies of 100, 200 and 239 participants. Larger samples let you see where views split across groups, and the people who disagree often hold the most useful information. They do not make individual answers more reliable, so keep the quality checks regardless.
Incentives
People give their time, and it is fair to pay for it. Hey Juno's standard is that incentives reflect the time and effort a study asks for, and that nothing is structured to pressure people into taking part or into saying what they think you want to hear.
Tell people up front what they will receive and when, and never tie the reward to a particular answer. Hey Juno prices incentives separately from the interview fee and shows participant payments separately. Participant incentives are in pilot and enabled per account, so ask before you plan around them. This guide suggests no amounts, because the right level depends on the audience, the length and the market.
Avoiding fraud and low-quality participants
Where a reward is offered, some people will try to collect it without giving real answers. Three layers help.
Before the interview. Use a screener with behavioural questions and decoy options. For panel recruits, use partners with their own identity checks. Duplicate-attempt controls reduce the chance of one person being counted twice.
During the interview. An interviewer that follows up on what someone says is harder to fool than a form, because specifics are harder to fake. If answers stay thin, Juno asks for more detail.
After the interview. Juno marks recruited interviews as Qualified, Disqualified or Incomplete, and shows the reasoning: the number of answers, whether short answers are excluded and whether personal context was given. You can adjust the rules per study. When some interviews fall short, Juno asks whether to report on all of them or only the Qualified ones. This is a different check from screening. Screening decides who enters the interview. Quality decides whether the result has enough effort to include.
Read a sample of transcripts yourself. No rule replaces that.
What this guide doesn't cover
This guide gives no incentive amounts, panel turnaround times or fees. It does not replace legal advice on consent in your market. Participant wellbeing has its own guide: participant wellbeing in AI-moderated research. The sample-size figures come from studies of small, focused samples and may not fit yours.
Frequently asked questions
Can I recruit participants for AI-moderated interviews from my own customer list?
Yes. Every Juno study has an invite link you can send yourself. Per-person links can carry customer record (CRM) context and interviewer notes.
Does going live in Juno start recruitment automatically?
No. Going live does not start recruitment. The Hey Juno panel is a managed service: a specialist scopes the study and recruits people who match your criteria. You can also share the link yourself or connect your own panel provider.
How many participants do I need for qualitative interviews?
In focused, uniform samples, published work suggests themes often settle within roughly 9 to 17 interviews, and understanding them fully often takes more. Plan for more to compare segments.
How do I stop fraudulent or low-effort participants?
Use behavioural screener questions, recruit through partners with identity checks, and review quality status afterwards. Juno marks recruited interviews Qualified, Disqualified or Incomplete, with the reasoning visible.
Do I have to pay participants?
No, but paying people for their time is fair, especially strangers recruited through a panel. Incentives are in pilot and enabled per account.