A second conversation is not a second opinion

The most encouraging user conversation we had this week was also the most misleading, and the two facts have the same cause. We talked again to someone who had been enthusiastic once before — a returning voice, warm, specific, full of detail about how he’d use the thing. It felt like validation. It was closer to the opposite: a reminder that a returning enthusiast is not new evidence, and that a signal can get louder without getting any truer.

Early customer development runs on a quiet accounting error. You count conversations when you should be counting distinct people. Talk to the same person twice and your notes double, your quotes pile up, your sense of “we keep hearing this” hardens — but the number of independent voices behind the claim is still one. The second conversation deepens the first; it does not corroborate it. Corroboration requires independence, and a person agreeing with himself is the least independent source there is. The trap is that depth feels like breadth: a rich, detailed second session reads as stronger evidence than a thin first one from a stranger, when for the question that actually matters — is this real for more than one person? — the stranger is worth more.

It got sharper than that. The pain he volunteered the second time was not the pain the product is built around. We build for one job; he described a different, adjacent one — easier for him to articulate, more concrete, and not the thing we’re betting on. So the warm returning signal was doubly hollow: not independent, and not even on-thesis. Filed carelessly, it would have gone into the ledger as “another data point for the wedge,” when it was neither another data point nor about the wedge.

The rule we’re keeping is boring and load-bearing: count distinct people, tag every signal with whose voice it is, and when the same person returns, file it as depth on an existing node, not a new one. It changes what you do next. A second session with a fan is not progress toward “is this real”; it’s progress toward “what exactly does this one person mean” — useful, but a different question with a different follow-up. The only way to move the first number is to go find a voice you haven’t heard yet. Enthusiasm is cheap to re-collect and expensive to mistake for reach.