60–90 minute calls do not scale
A proper JTBD interview is long, and you need dozens of them to see the patterns. Multiply 75 minutes by 30 customers across two time zones and the calendar problem alone kills the project.
Jobs-to-be-done without booking calls.
The JTBD interview is the gold standard for understanding why customers switch, and almost nobody actually runs them. Diaform is an AI interviewer that conducts the full jobs-to-be-done interview structure, probes the four forces, and captures the switch trigger moment, all over a shareable link. The depth of a moderated study, without the calendar.
14-day free trial · No demo required
An AI interviewer that runs the full JTBD interview structure
A real jobs-to-be-done interview is a 60–90 minute conversation that walks a customer back to the moment they switched, the trigger, the push from the old solution, the pull of the new one, the anxieties they had to overcome, and the habits holding them back. Done well, it is the single most predictive piece of customer research you can run. Done badly, it is a feature wishlist. The reason teams skip it is not because they doubt the method, it is because the method does not scale.
Diaform is the missing piece. The AI is trained on the JTBD interview structure. It opens with the timeline, walks the customer back to the switch event, surfaces the four forces by name without ever using the jargon, and presses on the desired outcome and the alternatives that were considered and rejected. Every session is asynchronous, every transcript comes back clustered into themes, and you finally get to run JTBD interviews at the scale the method deserves.
Why teams stop running JTBD interviews
Interviewer skill is the bottleneck
JTBD interviews live or die on the AI interviewer's ability to follow the timeline, probe the forces, and avoid leading questions. Most teams have one person who can do this, and that person is always busy.
Synthesis is days of transcript coding
Even when you get the interviews done, turning twenty hours of audio into push, pull, anxiety, and habit categories is a week of manual work. The insight arrives long after the decision was made.
How Diaform runs JTBD interviews end to end
Follows the JTBD interview structure
The AI interviewer opens with the timeline, walks the customer to the switch event, and stays in the structure even when the customer wants to wander into feature requests.
Probes the four forces by name
Push from the old solution, pull of the new, anxiety about switching, habit holding them back. The AI knows what to fish for, and what each force should sound like in a real answer.
Captures the switch trigger moment
The most important sentence in every JTBD interview is when the customer remembers the exact moment they decided. The AI is trained to recognise that moment and slow down to capture it in the customer's own words.
Theme clusters across interviews
Once you have ten or more sessions, the AI groups recurring pushes, pulls, anxieties, and habits across customers, with linked verbatim quotes, so the synthesis is done by the time the last interview ends.
Multilingual native interviews
The AI conducts the interview in the customer's native language and summarizes back to English for your team. Switching stories are deeply personal, running them in a second language loses the nuance.
No scheduling, ever
Share one link. Customers complete the JTBD interview when it suits them, on any device. You never open a calendar invite, and the response rate goes up, not down.
How to run a JTBD study with Diaform
- 1
Load your JTBD interview guide
Paste your existing guide, or start from a template. The AI follows your structure: timeline, switch event, four forces, desired outcome, alternatives considered.
- 2
Upload product and competitor context
Add notes on your product, the alternatives you compete against, and the language your customers use. The AI references this in real time when probing pulls and anxieties.
- 3
Share one link
Send the link to recent switchers, churned customers, or your target segment. Each respondent gets a fresh AI-led session in their own time, in their own language.
- 4
Review the synthesized JTBD insights
Each transcript comes back tagged by force, with the switch trigger highlighted. Across the cohort you get clustered themes, the dominant push and pull, and the verbatim quotes worth circulating.
What teams use JTBD interviews for
New product JTBD validation
Run JTBD interviews with the segment you think will switch to a new product, before you build. Find out whether the push is strong enough, or whether you are designing a vitamin nobody hires.
Competitive switching research
Interview customers who recently moved to or from a competitor. Surface the exact trigger, the anxieties they had, and the alternatives they almost picked instead.
Churned-customer JTBD post-mortem
Run a JTBD interview in reverse with churned customers. What pushed them away from you, what pulled them to the next solution, and what habit you failed to disrupt.
JTBD-driven persona work
Replace demographic personas with job-based ones. The AI surfaces the recurring jobs, contexts, and switching patterns across a segment so personas reflect behaviour rather than headshots.
Market entry research
Before entering a new market or vertical, run JTBD interviews with the segment in their native language. Find out what they currently hire, what they wish was different, and what would trigger a switch.
Feature deprecation research
Before killing a feature, interview the customers who use it about the underlying job. Decide whether the job goes away, gets folded in, or needs a new solution to take its place.
Traditional JTBD interviews vs. AI-led JTBD
Traditional moderated JTBD interviews
- 60–90 minute calls that do not scale beyond a handful
- Interviewer skill is the bottleneck on every study
- Weeks of manual transcript coding before insights land
- Single language, usually the researcher's, not the customer's
- Calendar friction sinks half the recruited respondents
Diaform
- 60-minute interview depth in an async format that scales
- AI follows the JTBD structure on every single session
- Theme clusters across interviews ready when the cohort closes
- Multilingual interviews in the customer's native language
- Switch-trigger probing built into every conversation
Frequently asked questions
Q
What is jobs-to-be-done?
Jobs-to-be-done (JTBD) is a research framework that explains customer behaviour through the lens of the job they are hiring a product to do. Instead of asking what features people want, a JTBD interview walks a customer back to the moment they switched solutions and surfaces the underlying job, the forces pushing them away from the old option, the forces pulling them to the new one, the anxieties about switching, and the habits holding them in place. It is widely considered the most predictive form of qualitative customer research.
Q
Can the AI really do a JTBD interview?
Yes. Diaform's moderator is built around the JTBD interview structure, the timeline, the switch event, the four forces, the desired outcome, and the alternatives considered. It probes for specifics, avoids leading questions, and recognises the trigger moment when a customer says it. It is not a replacement for a world-class researcher running a flagship study, but it consistently produces interviews that are deeper than what a junior researcher would run, at a scale neither could match.
Q
How long is each session?
Most JTBD sessions on Diaform run 25 to 45 minutes, shorter than a moderated call because there is no small talk, no rapport-building, and no calendar buffer. The AI keeps the conversation focused on the switch event and the four forces, and closes when it has what your guide asked for.
Q
Do I need a researcher to design the questions?
No. You can paste in an existing JTBD interview guide, or start from the built-in template that covers timeline, switch trigger, push, pull, anxiety, habit, desired outcome, and alternatives. If you have a researcher, they will get more out of it; if you do not, the structure is enforced for you.
Q
How many interviews should I run?
For a JTBD study, the patterns usually emerge between 12 and 20 interviews per segment. Because Diaform removes the scheduling and synthesis bottlenecks, most teams run 25 to 40, enough to see the dominant push and pull clearly, and to break the data down by sub-segment.
Explore more
Dive into the other ways Diaform can power your research.
Customer discovery
Run discovery interviews at the front of the funnel, pair them with JTBD interviews once you have switchers to talk to.
Persona research
Build job-based personas grounded in real switching behaviour rather than demographics and guesses.
AI user research platform
The full AI-led research stack behind Diaform, interviews, probing, synthesis, all from one shareable link.
Ready to upgrade your feedback loop?
Stop guessing why users leave. Start an automated interviewer in seconds and get the deep insights of a Zoom call at the scale of a survey.
14-day free trial · No demo required