Politeness is a strong force
When a person builds something and then asks you about it, the socially cheap response is encouragement. Discouraging someone requires effort, confrontation, and a willingness to be wrong in public.
Most people will not pay that cost for a stranger on a video call, so they hand you the polite answer and move on with their day. From your side, the polite answer is indistinguishable from a real one.
Prediction is unreliable, even honestly given
Ask someone what they'll do next quarter and they answer from an idealised model of themselves — the version with clear priorities, spare budget, and no competing fires.
That person does not exist in Q4. The answer wasn't a lie; it was a forecast made by an optimist with no accountability for the result.
The design fix
- 1Ask only about the past. 'The last time this happened, what did you do?'
- 2Ask for artefacts. 'Can you show me the spreadsheet?' A no is informative.
- 3Ask about cost already incurred. Money spent is the least deniable signal there is.
- 4Introduce a cost into the conversation. Ask for an introduction, a follow-up with their manager, a document. Willingness to spend a small amount of social capital predicts willingness to spend money.
- 5Never demo before you've finished listening. The demo converts a research call into a review.
Interpreting the twenty-six yeses
Go back through your notes and mark every positive response with what it cost the person to give it. Most will have cost nothing. Those aren't data points; they're conversational lubricant.
The handful that cost something — an introduction made, a document shared, a colleague pulled in — are your actual sample. Usually it is two or three people out of thirty, and those two or three are worth more than the rest combined.
The uncomfortable version
A research process that never produces a no is not a research process. If nobody has told you the problem isn't worth solving, you have been asking questions designed to prevent them from saying so.