Find out what your audience struggles with before you decide what to build. Open-ended interviews, nothing put in front of participants, and a report that shows you what you didn't think to ask about.
Problem discovery answers one question: what does this audience care about? Use it when you know who you want to serve but not yet which of their problems matter.
Candor interviews synthetic participants open-ended, with nothing put in front of them. The interviewer starts with who they are and how their days go, walks through a specific recent moment, then follows the friction: what they’ve tried and the workarounds they rely on. The report orders findings by how widely they came up, and shows whether participants raised each one unprompted or only when asked.
Run problem discovery when you know the audience and don’t know the problem: entering a new market, standing up a new product line, or picking up a segment your team has never spent time with.
It’s also the right call when you suspect your team’s picture of the customer is out of date. Discovery is where that picture gets checked, before it shapes a roadmap.
you already have specific problems in mind (that’s problem validation) or an idea for people to react to (that’s concept testing). Discovery is deliberately broad, so it won’t give you a verdict on any one hypothesis.
Nothing is shown to participants. The interviewer starts with who they are and how their days go, then follows the friction.
The setting they work or live in, often after a short warm-up on their role or their life.
A walk through a typical day or a specific recent moment, so answers come from something that happened rather than a general opinion.
A scan for friction, then a deep dive on what came up: what they’ve tried and the workarounds they rely on, and for business buyers, what’s at stake when it goes wrong.
For business buyers, who else feels the problem and who would need to care. For consumers, where it’s relevant, how the problem connects to the kind of person they are.
A close that can ask what they would change if they could, and who else would have strong views on it.
A report that separates what you went looking for from what you would have missed. Each section appears when the interviews give it something to show.
Discovery hands you a list of problems. Problem validation tells you which of them are real and painful enough to act on.
Use the themes to write a sharper guide for real interviews, and spend that budget on the threads that earned it.
Upload the transcripts. Participants are built from what you’ve already heard, so a synthetic round can go wide on what a handful of calls hinted at.
For the bigger picture, read where synthetic research fits in your workflow.
Problem discovery is exploratory. You don't know yet which problems matter, so nothing is put in front of participants and the interview follows what they raise. Problem validation starts from problems you already have in mind and tests each one, with a verdict per problem. They usually run in that order: discovery surfaces the candidates, then validation tells you which are real and painful enough to act on. If you already have a short list of problems you believe in, go straight to validation.
Because a problem people raise on their own is stronger evidence than one they agree with after you bring it up. The findings list is ordered by how widely each finding came up, and that ordering can flatter a problem that simply sounded reasonable when asked about. The marker shows whether participants raised a finding unprompted, only when probed, or a mix of both. When you choose what to validate next, check the markers before you trust the order.
You still get a check, just a different one. With assumptions entered, the report picks the most important assumptions behind the study, drawing on yours, and grades each as Supported, Partially supported, Contested or Not supported. Without them, it runs a framing check instead: it reads the assumptions built into the questions you asked, and never claims you believed them. Writing a few down is worth the minute it takes, because a contradicted belief is often the finding that changes the plan.
Yes. When you describe your audience you can upload interview transcripts, support tickets, survey data, market reports or internal briefs. Candor uses your documents ahead of web research, so participants reflect your customers rather than the broader population, and every signal drawn from a document names the file it came from. That makes discovery a good way to go wider on something a handful of real conversations hinted at.
No. It's the round before real interviews, or the round alongside them. Synthetic discovery is good at breadth: surfacing the themes worth exploring in a market you don't know well, early enough to shape the questions you take to real people. Real conversations are still where you confirm what matters most and hear the detail no evidence base holds. Most teams use discovery to decide what to ask real customers, not to skip asking them.
Start with a free project, or walk through Candor with us first.