AI moderated interviews, from first question to findings
In AI moderated interviews, an AI moderator conducts the conversation that a researcher would normally lead: it asks your questions, listens, and follows up on what the participant actually said. With Moderators, each participant opens your link, talks by chat or voice, and the finished interviews become themes with quotes. This page walks through exactly how one interview unfolds.
What decided it for you in the end?
The other vendor could start the same week.
Who needed it that week?
Finance did, because our launch date moved up.
Quote from the interview
“I knew it was in there somewhere, so I gave up and asked a colleague.”
Where does the first question come from?
From a guide you approve. You state what you want to learn and who you are talking to, and Moderators drafts goals, questions and suggested follow-ups. You can read it before anyone sees it. The guide is the plan, not a script. It tells the moderator what you need to understand.
How does it decide what to ask next?
It reads the last answer. A participant who says "the setup was a pain" gets asked what part, when it happened and what they did about it. A participant who gives a rich answer is not asked the same thing again. If a goal on your guide is still unanswered near the end, the moderator returns to it.
What is the difference for the participant?
Chat feels like messaging and suits short windows of time. Voice feels like a call with a patient interviewer and tends to bring out stories. Both follow the same guide and land in the same findings. Plans include a number of chat interviews and a number of voice minutes, so a team can offer either.
Is there a demo?
Yes. The home page runs a text interview: you enter what you want to learn and who you are talking to, answer three questions and see the first finding with a quote. Voice interviews and the full set of findings belong to paid plans.
What is stored?
A full transcript of each conversation. You can read any of them, search them and check any quote against what was said around it. Your data stays yours.
What do you receive across interviews?
Themes, ranked by how many conversations raised them, each with supporting quotes. The goal is that a stakeholder can read one page and see the evidence without hearing every recording. If you want to compare this against hiring a person, the post on AI moderator vs human moderator lays out the tradeoffs.
- Quotes link back to the transcript they came from.
- Themes reflect what was said, not what the guide hoped to hear.
- Unexpected topics that participants raised on their own are kept.
What should you not expect?
Moderators does not supply participants. You invite your own customers with a link, so the quality of a study depends on reaching the right people. It does not run a live group, and it does not record a screen. It is built for one-to-one conversation about experience, motivation and decisions.
When does it fit best?
When you need many conversations quickly and a human moderator would be the bottleneck. See the features overview for the full list of what the moderator does, or the user interview software page if you are comparing workflows.
What does an interview cost?
Annual billing is the default and takes 50% off. There is no free plan, and a used limit does not return within the period.
- Starter: $99/month, $49.50/month billed annually. 100 chat interviews, 120 voice minutes, 1 seat.
- Team: $249/month, $124.50/month billed annually. 400 chat interviews, 300 voice minutes, 5 seats, custom-branded link.
- Scale: $599/month, $299.50/month billed annually. 1,000 chat interviews, 750 voice minutes, 15 seats, findings repository, API.
- Enterprise: $1,499/month, $749.50/month billed annually. 2,500 chat interviews, 1,800 voice minutes, SSO, roles, SLA, invoices, data retention controls.
On annual Team billing, $124.50 divided by 400 chat interviews is about 31 cents per interview at the full limit. The pricing page has the rest.
Watch the moderator work on your own question
Enter what you want to learn, answer three questions and read the first finding.