Turn a belief into a decision
“Customers need a better solution” is too broad to test. A stronger hypothesis identifies a customer group, a recent problem, an existing alternative and an observable action. For example: maintenance managers at multi-site manufacturers will involve procurement in a pilot discussion because unplanned parts searches interrupt scheduled work. This is a hypothetical research hypothesis, not a market finding.
Before collecting evidence, specify the investment at stake. You may be deciding whether to fund a pilot, enter a segment or build a feature. The evidence needed for a reversible pilot differs from the evidence needed for a major production commitment. The SBA distinguishes existing market information from direct customer research; combine them only when they answer the actual question.
Build an assumption register
| Assumption | What would weaken it? | Next evidence |
|---|---|---|
| The problem is important | Buyers tolerate it without meaningful cost | Recent incident interviews |
| A budget can be accessed | No owner can authorize a purchase | Buying-process conversations |
| The offer fits the workflow | Adoption requires an impractical change | Observed task or limited pilot |
| The economics can work | Acquisition and delivery absorb the value | Explicit unit-economics scenario |
Rank assumptions by decision impact and current uncertainty. Test the one most capable of invalidating the investment, even if another question is easier or more comfortable to investigate.
Match the test to the claim
Interviews can explain language, tradeoffs and buying mechanisms. A survey can measure responses within a defined sample, subject to its design and recruitment limits. A pilot can show behavior in a constrained setting. A landing-page response can show interest in a message, but not establish that the organization can buy and adopt the product.
Choose a decision rule before viewing the results. There is no universal number of interviews or responses that proves a market. Explain what counts as sufficient evidence for this particular next step and which uncertainties will remain.
Keep negative cases visible
Record contradictory findings with their source, role and context. A technical user may value the offer while an economic buyer sees no reason to change. Treating both as a single average opinion can hide the adoption problem. Look for the conditions under which the hypothesis holds.
Write a decision memo
Summarize the hypothesis, evidence gathered, recruitment limits, competing explanations and recommendation. Choose proceed, revise, pause or stop, with an owner and next review. A useful validation result can be a smaller investment or a different audience. Validation should reduce uncertainty; it should not become a way to decorate a decision that has already been made.