Hey Juno

Informed consent in customer research interviews

By Hey Juno · Published · Updated

Informed consent in customer research interviews means a participant knows who is asking, why, how their answers will be used and kept, and that they can skip questions or stop, before they agree to take part. In an AI-moderated interview it also means saying plainly that an AI conducts the conversation. This guide covers what to tell people, how to handle recording and quotes, and what to check with your own legal or privacy team.

What should participants be told before an interview?

Tell people enough that their "yes" is a real choice. For qualitative research, that usually comes down to six things:

  1. Who is asking. The organisation commissioning the study, and the organisation running it if they differ. People answer differently when they know whether they are talking to a brand, an agency or an independent body.

  2. Why. What the research is for, in a sentence a stranger would understand. Avoid a purpose so vague that anything could fit inside it.

  3. That an AI conducts the interview. If Juno, the AI researcher from Hey Juno, or any other AI is interviewing, say so at the start. Do not leave people to find out halfway through. See what taking part in an AI interview is like for how this looks from the participant's side.

  4. How answers are used and kept. Who will see the transcript or recording, whether quotes may be shared, and how long the data is held.

  5. That they can skip or stop. No question is compulsory, and they can leave at any point without explaining.

  6. What they receive. Any incentive, and what happens to it if they stop early.

Plain wording matters more than legal completeness. A short, readable introduction that people actually read beats a long form they scroll past. Put the long version behind a link for those who want it.

What does Juno do in its participant flow?

Juno's participant flow starts with an introduction and consent step. Everyone who opens a study link completes the study's introduction, consent and any pre-interview questions before the interview begins. Customers can set their own brand for the participant experience.

Hey Juno's published participant standards describe consent as telling people what the research is for and how their answers will be used before they take part. They also commit to purpose limitation: answers are used for the research they were given for, not quietly repurposed. Participants can withdraw, and can ask to access, correct or delete what they shared.

You can see the introduction yourself. In Rehearse you can take the interview yourself, so you can check the introduction before any real person sees it. A consent screen you have never read is one you cannot vouch for.

Consent to recording and to quotes

Agreeing to an interview is not the same as agreeing to be recorded, or to be quoted. Keep them separate in your own mind, even if they sit on one screen.

  • Recording. Say whether the session is audio, text or both, and who can hear or read it. In Juno studies, customers hear the original recording. Juno does not redact transcripts, quotes or exports today. Automatic redaction of hard identifiers is planned but not yet switched on, and even when it is, it will not be anonymisation: details such as an employer and a city can still point to a person. Access is limited to people in your organisation who can see the study. Do not tell participants their answers are anonymous unless you have checked that this is true.

  • Quotes. If quotes may appear in a report, a deck or a pitch, say so. State whether they will carry a name, a first name, a role or nothing. Participants decide more freely when they know what attribution to expect.

  • Sensitive topics. Health, money and family deserve a clearer notice and tighter handling. Hey Juno's standards describe tighter handling and clearer retention for these categories, and data minimisation as the default: capture what the study needs and no more.

Retention is the part teams most often leave vague. Decide how long you keep data and tell people. Hey Juno lets the customer set retention and request deletion.

Minors and vulnerable people

Handle this with care and take advice. In general terms:

  • Children and young people usually need consent from a parent or guardian, and the age at which a young person can consent for themselves varies by country. Check the rules and codes that apply where your participants live.

  • Some adults may be less able to give free consent: people in distress, people with certain cognitive conditions, people who depend on the organisation asking. Consider whether the study should reach them at all, and what support to give if it does.

  • If a study is not designed for these groups, say so in screening and make it easy to leave.

Participant wellbeing during the interview is a separate subject. Juno applies a live, four-tier wellbeing framework in every interview. For how it works, read wellbeing in AI research and the safety standards rather than relying on this guide.

Incentives and pressure

An incentive should recognise a person's time. It should not push them to say yes, or to say what you hope to hear. Hey Juno's standards say incentives reflect the time and effort a study asks for, and that nothing is structured to pressure people into taking part or into telling Juno what it wants to hear.

Practical checks:

  • Say the incentive up front, and what happens if someone stops early.

  • Do not tie payment to a particular answer or to finishing every question.

  • Be careful where there is a power gap, for example interviewing your own customers who also depend on you for service. Make clear that declining has no effect on their account.

Can participants withdraw?

Yes, and they should know how. Withdrawal has two parts: leaving the interview, and removing what was already collected. Tell people both are possible, how to ask, and what has to happen first. For example, if findings have already been published in aggregate, data may be harder to separate. Say that honestly instead of promising more than you can deliver.

If someone stops partway, you can use what they gave before stopping only if they agreed to that. If they ask for deletion, delete it, and make sure the request also reaches any exports you made.

What do the industry codes say?

Two codes are worth reading in full:

Neither is a template for your consent text, and this guide does not summarise their detailed rules. Read the current versions and follow the ones that apply to you. Data protection law may also apply. The UK ICO's research provisions guidance explains that agreeing to take part is not always the same as consent as a legal basis for processing. Check with your own legal or privacy team.

Where this doesn't apply

This guide is general information about good practice, not legal advice. It does not set out the rules of any one country, and it does not cover clinical research, which has its own ethics approval and consent requirements. Rules for children, for health data and for public bodies differ by place. Simulated interviews used for rehearsal involve no real participants, so none of this applies to them. They are never research data.

Frequently asked questions

Do participants need to be told an AI is interviewing them?

Yes, as a matter of good practice, and some places have transparency rules about it. Say it at the start in plain words. See AI-moderated vs human-moderated interviews.

Is a tick box enough?

A tick box records agreement. It does not create understanding. Short, specific wording in the introduction does more than a long form. Test it by asking someone outside the project to explain back what they agreed to.

Can I quote participants if they agreed to take part?

Only if you told them quotes might be used and how they would be attributed. Agreeing to an interview does not by itself cover publication. Be specific, and take extra care with distinctive details that could identify someone.

What if someone wants their answers removed after the interview?

Treat it as a request to act on promptly. Remove their data from the study and from any exports you hold, and tell them what was done. If something cannot be undone, such as figures already published in aggregate, say so plainly.