Problem validation

When to run problem validation

Problem validation is the right call when you have specific hypotheses about what your audience cares about and you need to know which to invest in. Hypotheses come from anywhere: a prior discovery study, customer support tickets, sales calls, a leadership-team workshop, an executive’s strong intuition. Wherever they come from, the next question is the same. Are these problems real, do the right segments feel them, and which ones are worth solving?

Without validation, teams often invest engineering effort into problems that turn out to be edge cases or executive pet theories. The cost shows up later, when a feature ships and the metrics don’t move. Validation early is the cheapest filter you have.

How Candor runs it

You state the hypotheses up front. Candor runs structured interviews probing each hypothesis against the persona’s actual situation, with follow-up questions that test the persona’s reaction in the context of their work. Hypothesis-introduction questions are delivered verbatim from your study guide so the framing stays consistent across every persona, and the interviewer agent then probes for behavioral evidence, frequency, severity, and counterfactuals.

The critic agent reviews every persona response against the persona’s prior statements and the underlying evidence. If a persona contradicts itself, or if a response drifts from what the source evidence supports, the response is regenerated before you see it. You get clean signal, not the noise of an inconsistent persona accumulating across a long study.

What you walk away with

A synthesis report with explicit verdicts on each hypothesis: supported, contested, or not supported. Each verdict comes with the supporting and contradicting evidence, the archetypes that confirm or deny it, and the specific persona quotes behind the call. Contested hypotheses are tagged with which segments confirmed and which didn’t, so you can decide whether you’re looking at a real problem with a narrower audience than you thought, or a problem that doesn’t hold up.

The output is designed to be acted on. Killed hypotheses stop consuming roadmap time. Validated hypotheses move forward with evidence behind them. Contested hypotheses become segmentation decisions. The report links every claim back to the source quote, so a stakeholder review can audit any finding without a handoff document.

Where to go next

If you don’t yet have hypotheses to validate, start with problem discovery to surface them. Once you have validated problems, move to concept testing to pressure-test candidate solutions, or value-prop testing to refine how you talk about the problem and the solution. To stress-test the broader assumptions sitting behind the work, run assumption validation.