Customer enthusiasm is not evidence.

The first customer conversations are not a survey and they are not a pitch competition. Their job is to expose where the founder’s mental model is wrong: who owns the pain, how the problem is handled today, what makes it urgent, and what would have to change for someone to buy.

Start with behaviour, not opinions

“Would you use this?” invites politeness. Ask what happened the last time the problem occurred, who was involved, what it cost, which workaround was used and what happened next. Past behaviour is imperfect, but usually more informative than hypothetical enthusiasm.

Interview the buying system

In B2B, the user, champion, budget owner, security reviewer and procurement team may all be different people. Ten conversations with enthusiastic end users can still leave you with no path to purchase. Map who experiences the problem and who can authorise change.

In B2B, the user, champion, budget owner, security reviewer and procurement team may all be different people.

Look for repeated language and repeated cost

A pattern becomes meaningful when different customers describe the same painful moment in their own words and attach a real consequence to it: revenue leakage, headcount, risk, delay, customer churn or missed capacity. Repeated pain is more useful than repeated compliments.

Let the thesis move

The conversation program has failed if the founder finishes with exactly the same proposition, customer and workflow they started with. Strong discovery should sharpen something: segment, urgency, buyer, scope, pricing logic or the product itself.

Design the conversation sequence

The first ten conversations should not all be identical. Early calls can map the problem and vocabulary. The next set should test the sharpest emerging segment, urgency and buying system. Later conversations should ask for progressively stronger commitments: a second meeting with the budget owner, access to workflow data, a prototype session, a pilot or a commercial discussion. If the questions never become harder, the evidence is unlikely to become stronger.

Common failure mode: interviewing only friendly users

Warm contacts are useful for access, but they can create false confidence. Include people who do not know the founder well, customers with different levels of urgency and people involved in procurement or risk. The aim is not a representative market survey. It is to expose where the proposition breaks before product and sales investment make those assumptions expensive.

Signal to watch: the quality of conversations should improve as the thesis sharpens. If the tenth interview is still producing only broad statements such as “interesting” or “useful,” change the questions, the segment or the requested commitment rather than simply increasing the sample size.

Applied example

A founder hears from ten HR leaders that workforce planning is “important.” That is weak evidence. A stronger pattern appears when six describe the same quarterly spreadsheet process, three say it delays hiring decisions by weeks, and two agree to share anonymised data for a pilot. The conversations have moved from opinion to behaviour and commitment.