Customer interview mistakes in recruiting and booking

The mistakes that ruin an interview are mostly not made in the interview. They are made when choosing who to call, and again in the days around the call, when it is booked, recorded and opened.

Each mistake below is paired with what it does to the answers and the smaller choice that replaces it. They follow the order in which a founder meets them: recruiting, booking, the first minute, and the last five.

What to take away

Recruiting: who gets the invitation

Screening for availability instead of relevance. You call the people who answer. Availability correlates with willingness to help you and not at all with having the problem. The screener design turns relevance into factual questions you can check.

Explaining the problem in the invitation. By the time they arrive they know the right answer and have had time to prepare a helpful version of themselves. Invite them to talk about how they currently do something, not about your topic.

Paying in a way that recruits professionals. An incentive attracts people who take incentives. If you offer one, keep it small, keep the screening strict, and note in your log which conversations were paid.

Talking only to current users of the thing you want to replace. They already selected into a solution. The people who looked and did not buy are the more informative group, and they are much harder to find, which is why nobody talks to them.

Run the checklist once more before the invitations go out.

The disclosure that paid conversations carry

A payment is a connection between you and the speaker, and that note matters again later. If you ever quote the person publicly, the payment has to be disclosed.

The FTC's endorsement guidance says a connection that consumers would not expect, and that would affect how they weigh the statement, should be disclosed clearly and conspicuously. A log line written on the day of the call makes that disclosure a copy-and-paste job later.

Booking: ten in a week, two colleagues, one recording

Scheduling ten in a week. You will run the last six with the conclusions from the first four already in your head, and you will hear confirmation. Space them so there is time to write up between.

Bringing two colleagues. Two people on your side turns a conversation into an audience, and the answers get shorter and more polite. One interviewer, one silent note-taker at most.

Recording without asking properly. Ask before the call, not as the recording starts, and say who will hear it and how long you will keep it. Someone who agreed under pressure will be careful for the rest of the conversation.

The first minute

Opening with your credentials. A minute spent establishing why you are worth talking to sets up an expectation that you know things, and people do not contradict experts. Open with a question about last week.

Accepting a hypothetical. Every question about would, could or might returns a guess. Every question about last time returns an account. This is the highest-yield habit in the whole practice, and it is not a quirk of founders.

Even professional survey designers lean on testing for this. The Census Bureau's questionnaire standards say pretesting is critical to finding problems with question content, which is why a first call is worth treating as a test of your own questions. The interview mechanics cover the moves inside the room.

The last five minutes: recruiting the next round

Ending on enthusiasm. Warmth at the end of a call costs nothing, and it is what you will remember. Make it slightly expensive instead: ask for the spreadsheet they described, an introduction to the person who signs, or half an hour to watch them work.

Not asking who else. The next name is the cheapest recruiting you will ever do, and it is the one route that repeats without a budget.

Leaving the buyer unnamed. Ask who signed off the last comparable purchase before the call ends. It is a factual question about the past, and people answer it.

Keep the recruiting trail in the log

Each row of the log should say how you reached the person, whether they were paid, and whether they agreed to a recording. Those three facts are what let you judge, later, whether a batch described a market or only your address book.

Write up the same day. A conversation summarized a week later has already been edited toward what you wanted to hear, and memory fades toward whatever hypothesis you are carrying.

When you count, count workarounds, existing spend and who could authorize a purchase, as in the discovery counts. Tallying how many people liked the idea produces a number with no unit.

Common questions

How many people should be in a round?

Enough that you stop being surprised, which depends on how varied they are. Run them in small batches with a write-up between, and stop when a batch changes nothing in your written problem statement.

Is it a mistake to interview someone twice?

No, and it is underused. A second conversation months later, about what they actually did in between, is closer to evidence than any first conversation can be.

What if they ask what I am building?

Tell them at the end. Say at the start that you would rather hear about their situation first and will happily explain afterwards. Almost nobody objects, and the answers before the reveal are worth keeping separately from the ones after.

Do I have to disclose that someone was paid?

If you ever quote the person publicly, yes, because the payment is a connection that has to be disclosed. Keep a note of which conversations were paid so the answer is already on file.