Why most founder interviews fail
The problem is almost never that founders don't talk to customers. It's that the conversation is built out of questions people can answer without having done anything.
Ask someone whether a problem is frustrating and they will say yes, because saying no would be rude and because most things are mildly frustrating. Ask them what they did the last time it happened and you get either a story or a silence. Both are useful. The first is evidence; the second is your answer.
The whole discipline is this: ask about the past, not the future. Memory is imperfect but it is at least a record of something. Prediction is just imagination with a confident voice.
Eight questions to stop asking
- —Would you use this? — Predicts nothing. Costs nothing to say yes.
- —Would you pay for this? — Hypothetical money is infinitely available.
- —Do you think this is a good idea? — Invites them to review you, not the problem.
- —How much would you pay? — People invent a number to be helpful. It is not a price.
- —Does this happen often? — 'Often' means something different to each person. Ask for the last time instead.
- —What features would you want? — Turns a customer into a junior product manager.
- —Is this a big problem for you? — Every problem is big while you are discussing it.
- —If we built X, would you switch? — Switching costs are invisible in a hypothetical.
Opening: establishing the context
- 1Walk me through the last time you had to do this. Start from the moment you realised you had to.
- 2Who else was involved? What did each of them do?
- 3How long did the whole thing take, start to finish?
- 4What were you using to do it? Name the actual tools, including the spreadsheet.
- 5When was that — this week, last month, last quarter?
- 6How did you know it was time to deal with it? What triggered it?
Middle: finding the friction
- 1Where did it go wrong, if anywhere?
- 2What did you do when it went wrong?
- 3Has that happened before? Tell me about the previous time.
- 4What did you try before you settled on doing it this way?
- 5Why did you stop using that?
- 6What's the part of this you'd hand to someone else tomorrow if you could?
- 7Who ends up chasing you about this, and what do they say?
- 8What happens if it just doesn't get done for a week?
Economics: is anyone funding this?
- 1Has anyone spent money on this problem already? What on?
- 2Who signed off on that? What did they need to see?
- 3What's the budget line this would come out of, if there is one?
- 4How much of your week does this take? Whose week?
- 5Has this ever been raised in a planning conversation? What happened?
- 6If you tried to get this bought, who would say no first, and why?
Language: how they actually say it
This is the part most founders skip, and it's the part that changes your homepage. You are listening for the words that come out unprompted — the phrases they'd use to describe the problem to a colleague.
- 1What do you call this internally?
- 2If you were explaining this to a new hire, how would you describe it?
- 3What would you type into a search bar if you went looking for a fix?
- 4Is there a name for the role of the person who owns this?
- 5When you complain about this, what do you say?
Closing: the questions that surface everything else
- 1What haven't I asked about that I should have?
- 2Who else deals with this that I should talk to? Would you introduce me?
- 3What would have to be true for you to change how you handle this?
- 4Is there anything about this conversation you'd want to correct after thinking about it?
- 5Can I come back to you in a few weeks with what I've learned?
How to run the conversation
Aim for the customer talking about eighty percent of the time. If you are talking more than that, you're pitching.
When you hear something interesting, do not move on. Ask 'what happened next' until the story runs out. The useful detail is almost always three follow-ups deep.
Silence is a tool. Count to four after they finish an answer. People fill the gap with the thing they were hesitant to say.
Record it if they consent, and keep the transcript. The exact wording matters more later than your notes will suggest at the time.
Do not demo. If you demo, the rest of the conversation is a review of your product rather than a record of their life.
What to do with the answers
Split every answer into two columns: things you observed, and things you were told about the future. Only the first column is evidence.
Then count. How many of the people you spoke to have already spent money, time, or political capital on this problem? If the answer is none, you have interest, not a market.
If you want to test what you've gathered against a scoring system, the Kill Decision assessment does exactly that in ten questions.