USE CASE

Problem discovery

Surface the pains, frustrations, and unmet needs in your audience before you scope a solution. Open-ended interviews with synthetic users grounded in real evidence, who remember their own context and push back when something doesn’t fit.

When to run problem discovery

Problem discovery is the right starting point when you know who you want to serve but you don’t yet know what problems matter most to them. It’s the work that comes before solutioning. You’re not testing an idea; you’re building the map of where the pain lives.

The most common moments for problem discovery: entering a new segment, market, or geography; standing up a new product line; redirecting a roadmap after a strategy shift; replacing intuition with evidence after a leadership change; setting up a research program at a company that’s never had one. In each case the risk is the same: building a solution to a problem your audience doesn’t actually have.

How Candor runs it

You define the audience and upload any prior research you have. Candor builds a synthetic audience grounded in that evidence, generates archetypes, and materialises individual personas with personalities, biases, and memory. You interview them in open-ended conversation, either live or through the auto-interview agent, with a guide that probes their daily work, their constraints, what they’ve tried, and where they’ve hit walls.

Personas remember everything across sessions, so you can return to a persona to probe deeper without re-establishing context. The critic agent validates each response against the persona’s prior statements and the underlying evidence, so you don’t silently accumulate AI agreeableness drift across a long interview.

What you walk away with

A synthesis report that surfaces tagged signals (pain, behavior, belief, goal, blocker), themes clustered by frequency and intensity, archetype-specific breakdowns showing how different persona types responded differently, and the genuine tensions where personas disagree. Tensions are usually the most valuable output. They’re where two segments need different things, and that’s where positioning, segmentation, and roadmap decisions actually live.

Every signal in the report links back to the specific interview quote that produced it, and every quote is attached to the persona who said it, with the evidence sources that grounded that persona’s view. You can audit any finding back to its origin.

Where to go next

Most teams follow problem discovery with problem validation on the two or three problems that look most worth investing in. From there, a candidate solution is best pressure-tested with concept testing. If you want to stress-test the strategic assumptions sitting behind the work, layer in assumption validation.

Common questions

Problem discovery is exploratory. You don't yet know which problems matter; you're surfacing them. Problem validation is hypothesis-driven. You already have specific problems in mind and you're testing whether the evidence supports them. The two often run in sequence: discover, then validate the discoveries you want to invest in.

A typical discovery study runs across 8 to 16 personas spread over multiple archetypes. The exact number depends on how heterogeneous your audience is. If your audience splits cleanly into two or three buyer types, fewer personas per archetype works. If your audience is broad (multiple industries, multiple roles, multiple geographies), generate more personas to cover the spread. The synthesis surfaces patterns that hold across archetypes and tensions that don't.

A good problem signal is specific, behavioral, and recurrent. Specific means the persona names a concrete activity, situation, or constraint, not a vague feeling. Behavioral means it shows up in what the persona does, not just what they say they want. Recurrent means it appears across multiple personas or archetypes. Candor tags signals by type (pain, behavior, belief, goal, blocker) so you can filter for the kinds you want and see frequency by archetype.

Yes. Upload PDFs, transcripts, survey data, market reports, or anything else you've already gathered. Candor uses your research as the primary evidence source for the synthetic audience, and every persona attribute that derives from your documents is tagged with provenance. Discovery interviews then layer on top of the audience that emerged from your data, not a generic audience built from web research alone.

Synthetic problem discovery is fastest when you don't yet have access to real customers, when recruitment is slow or expensive, or when you're entering a market you don't have relationships in yet. It's also useful for generating hypotheses you'll later test with real customers. The pattern most teams converge on: synthetic for breadth and hypothesis generation, real customers for depth on the few hypotheses that matter most.

More FAQs →

Candor is in development.

Be the first to know when it launches.

No spam. Just a note when Candor is ready. Powered by Highline Beta.