Product discovery reduces uncertainty about what to build. Start with the customer problem and the outcome you want, then identify what must be true for your idea to work. A request for a dashboard may conceal a problem that a simpler notification would solve.
Four useful risks are value (will people choose it?), usability (can they use it?), feasibility (can we build it?) and business viability (can the business support it?). They overlap, but naming the risk helps you choose evidence.
Use the cheapest responsible test that answers the question. A prototype can test comprehension without production code; a technical spike can examine a difficult integration. State a decision rule before the result arrives. A positive interview alone does not prove people will change their behavior.
The test follows the assumption, not the preferred solution.