Customer interview mistakes: 5 false positives

The most expensive customer interview mistakes don't feel like mistakes: the calls are friendly, people like the idea, and the notes say "positive". An interview supports an idea only when it contains facts about what the person has already done, spent or committed to; praise, promises and requests are false positives. A false positive is a signal that convinces you the idea works when it doesn't, and it is worse than no data, because it adds confidence. Below are five kinds of false positive, what to ask after each one, and a way to score an interview afterwards.

Twelve fans, two subscribers

CatChow, an online pet-food shop, is considering a subscription: cat food that arrives before the bag runs out. Sofia, a junior product manager, talks to 14 cat owners in three weeks and posts a summary in the team chat: "12 of 14 are positive. People love it."

Six weeks later a pilot opens, and Sofia writes to each of the twelve fans personally. Two subscribe.

The ten who didn't subscribe weren't dishonest. Most were polite, optimistic or trying to help, and the questions made that easy. The interviews never measured demand.

Five false positives and what to do with each

1. Compliments

How it sounds: "Great idea!", "I love it."

Why people say it: kindness is free. Even a sincere compliment is an opinion about the future, so it is a guess.

What to do: say thanks, write nothing down and go back to the person's life: "How do you handle this today?"

2. Fluff

How it sounds: a generic claim ("I usually buy ahead"), a future promise ("I would subscribe") or a hypothetical maybe ("I might order treats too").

Why people say it: they describe their best selves and best intentions. It isn't lying, just blurry: no date, no place, no money.

What to do: anchor the fluff to one real event: "When did you last buy food? Walk me through it." For a promise about a product that doesn't exist yet, anchor what does exist: what they pay now, whether they have ever switched shops.

3. Feature requests taken at face value

How it sounds: "You should add reminders to the app."

Why people say it: they want to help, and proposing a fix is easier than describing a problem.

What to do: write it down, keep it off the roadmap and dig for the reason: "Why do you want that?", "What would it let you do?", "How do you manage without it now?"

4. Polite non-commitment

How it sounds: "Keep me posted", "Send me the deck."

Why people say it: refusing feels rude, and these phrases end the conversation without costing anything.

What to do: propose one concrete step that costs the person time, reputation or money, the three commitment currencies: "Pre-orders are open. Shall I send you the link?" Someone who stays friendly but never commits or refuses is a zombie lead, and a pipeline full of them inflates the forecast.

5. Your own part

How it sounds: this time it's you. Pitching: "Let me explain how it works." A leading question: "It's annoying when the food runs out, isn't it?" A guest list of friends, colleagues and loyal fans.

Why you do it: you want the idea to survive the conversation.

What to do: bring up the idea last or not at all, talk less than the other person and recruit people who owe you nothing. For ready-made wording, see these customer interview questions.

What you heard, what it means, what to ask next

What you heard	What it actually means	What to ask next
"Great idea, I love it!"	The person is being kind	"How do you deal with this today?"
"I usually buy in advance"	This is who they would like to be	"When did you last buy food? Walk me through it"
"I would definitely subscribe"	An optimistic guess about their future self	"What do you pay now? Have you ever switched shops?"
"You should add reminders"	Something isn't working; you don't know what	"What would reminders let you do? How do you manage now?"
"Keep me posted"	A polite exit	"Pre-orders close on Friday. Shall I send the link?"
"Yes, that's annoying" (after a leading question)	You wrote the answer yourself	"Tell me about the last time it happened"

One call, line by line

Sofia's call with Grace, one of the twelve fans:

Who	Line	What is happening
Sofia	"We're building a subscription so food arrives before you run out. What do you think?"	A pitch, then an opinion question. Only one answer is polite
Grace	"Oh, great idea! I'd definitely use it."	A compliment plus a future promise. Both are free
Sofia	"It's annoying when the bag is suddenly empty, right?"	A leading question: the answer is inside it
Grace	"Yes, it usually happens at the worst moment."	A generic claim. No date, no place, no story
Grace	"You should add reminders to the app."	A feature request: a clue, not a spec
Sofia	"Good one, I'll add it to the list."	Taken at face value. Nobody asked why
Sofia	"Can I count you in at launch?"	A request for a promise, not for a commitment
Grace	"Sure, keep me posted!"	A polite exit. Nothing is agreed

Sofia's note after the call: "Grace: big fan, wants reminders, will subscribe." Eight lines, zero facts.

Later Sofia calls Grace back with three questions:

"When did you last buy cat food?" "On Saturday, at the supermarket, with the weekly groceries."
"When did you last run out?" "I can't remember. Not this year."
"What would reminders let you do?" "That's for the flea drops. They're monthly and I keep forgetting. Last month I was two weeks late."

Grace has no problem with food; her problem is a monthly treatment. Bad news for the subscription, and the first real data from this call.

Score the interview afterwards

Right after each conversation, answer three questions from your notes. Each "yes" is one point.

Facts about the past. Is there at least one specific event: what the person did, when and where?
Money or time already spent. Have they paid, searched, switched or built a workaround because of this problem?
A commitment given. Did they give you time, reputation or money: a booked follow-up with a purpose, an introduction, a pre-order?

Separately, mark the direction: do the facts support the idea or contradict it? The score says how much the interview can be trusted, not whether the answer was yes.

Sofia's three weeks, rescored from the original notes:

Interviews	What the notes contain	Past facts	Already spent	Commitment	Score each
4	Praise only	0	0	0	0
3	"I would subscribe", "I usually run out", "keep me posted"	0	0	0	0
1	Praise and a feature request nobody dug into (Grace)	0	0	0	0
2	Ran out of food this year and paid for a same-day courier	1	1	0	2
2	Order from two shops, keep a calendar reminder, booked a follow-up call	1	1	1	3
2	Buy at a shop on their usual route, have never run out	1	0	0	1

Eight of the 14 interviews score zero. They aren't positive, they are unknown. Four describe a problem that people already pay to solve, two show no problem, and only two contain a commitment: they came from the two people who later subscribed.

The honest report is no longer "12 of 14 positive". It is: "14 interviews. Four people have the problem and already spend money on it; two of them booked a follow-up, the other two were never asked for anything. Two have no problem. Eight are unknown, and we are calling them back."

After the callback, Grace's interview moves from 0 to 1 point and counts against the idea.

Common mistakes when you start filtering

Counting heads. Fourteen interviews are not a survey, and "12 of 14" is not a conversion rate.
Interrogating. Ask about the last time with curiosity, not suspicion.
Asking for a commitment too early. First learn how the person lives with the problem, then make the offer.
Treating "no problem" as a failed interview. It is a result, and cheaper than a failed launch.

What to do in practice

Before: decide which commitment you will ask for.
During: after praise, a promise or a request, ask about the last real event.
At the end: propose a next step with a who, a what and a when, and make "no" easy to say.
After: score the interview and report facts, spending and commitments, not moods.

Interviews are one step of a longer process: see how to validate a product idea.

Key takeaways

Compliments, fluff, feature requests and "keep me posted" feel like validation and cost the speaker nothing. Your own pitching invites them.
The cure is always a question about what already happened.
Score each interview on past facts, money or time already spent, and commitments given.
An interview with no facts is unknown, not positive.

FAQ

What is a false positive in a customer interview?

Anything that makes the idea look wanted without showing that it is: praise, "I would buy it", an unexplained feature request, a friendly "keep me posted".

Should I never mention my idea?

Early on, keep it out: you are learning how people live with the problem. Later you have to show it, and compliments will come back. That is the moment to ask for a commitment.

Does a low score mean the idea is bad?

No. Zero points mean the interview holds no evidence either way, so call back and ask about the last time. The idea is in trouble when interviews full of facts keep showing no problem and no spending.

Practise spotting them

In the free lesson Compliments and how to deflect them you practise on CatChow transcripts: spotting praise in notes, steering back to facts and rewriting an opener that fishes for approval. It is part of the free course Customer Development: Talking to Customers, built around questions that get honest answers.

Customer interview mistakes: 5 false positives that "validate" weak ideas