Customer interviews that don't lie to you

Two conversations, booked this week

Let's name the thing directly: this is the lesson people stall on. Reading about interviews is comfortable. Asking a near-stranger for twenty minutes of their time is not, and so students park it, and a week becomes a month.

So here is the deal, stated up front. Two interviews are required to pass the Module 2 Project. Three are recommended. Two is the number at which a second voice can confirm or contradict the first — the minimum for evidence rather than anecdote. Three lets patterns start to show, which is why it earns more in the project rubric. But we will not hold your progress hostage to a third stranger's calendar, and strong desk research and survey data can carry part of the weight. More on that in Lesson 2.6.

The rest of this lesson exists to remove every excuse between you and two booked conversations. Who to ask. The exact message to send. What to do when someone says no. And how to run twenty minutes without accidentally poisoning your own evidence.

People politely lie — unless you ask about the past

Ask someone "would you use this?" and they will look at your hopeful face and say yes. Not because they're dishonest — because they're kind, and because the future costs them nothing. Rob Fitzpatrick's The Mom Test named the fix, and it is the cardinal rule of this lesson:

Ask about what they did, never about what they would do.

"Would you book physiotherapy online?" produces an opinion about an imaginary future. "How did you book your last physio appointment — walk me through it?" produces facts: the phone call at lunch, the receptionist's voicemail, the two-day wait. Facts about the past are the only currency in a research interview. Opinions about the future are counterfeit, and compliments are worse — "that sounds great, I'd definitely buy that" is politeness data, worth nothing.

The second rule follows: never pitch. The moment you start selling, your informant becomes an audience and starts being kind. If they ask "is this about your product?", say you're only trying to understand how they handle the problem today, and park the product for a different conversation. Discovery and selling never share a call.

Mateus's first interview, rescued

Mateus at Ostara Ceramics tried this on a friend of a friend. His opening question: "Would you pay €78 for a handmade tableware set?" The answer — "sure!" — meant nothing. It cost her nothing to say and committed her to nothing.

Then he rescued it: "When did you last actually buy tableware? Walk me through it."

Out came a story containing four usable facts. It was a wedding gift, not a purchase for herself. She found the maker through a gift-guide blog, not Instagram. She chose it because "it didn't look like everyone's Instagram plates". And she almost didn't order, because she was worried the ceramics would break in shipping.

One rephrased question. Four facts that Ostara's marketing had never surfaced — including a buying occasion, a discovery channel and an objection. That is the exchange rate this lesson is offering you.

Who to ask — including when you have no customers

You want people with recent, real behaviour to describe. Three groups, in order of value:

• Recent buyers — yours, if you have them; a competitor's if you don't. Someone who bought from a rival last month is a perfect informant.
• Recent almost-buyers — people who looked, compared, and didn't buy. They know where the anxiety lives.
• Recent leavers — customers who stopped. Nobody else will tell you the truth they carry.

"I don't have customers yet" is therefore not a blocker. The behaviour you're studying exists whether or not you were the one selling — interview people who recently bought from anyone in your category, or who recently had the problem and solved it some other way, including by ignoring it. Sandbox students do exactly this: any accountant you can reach is a valid informant for Storkflow, any recent gift-buyer for Ostara, anyone who booked physiotherapy this year for Herzog. The business is fictional; the interviewing skill is not.

Recruit from your own customer records, from your network's second degree ("who do you know who recently…?"), or from communities you already legitimately belong to. Recency beats seniority every time: one person who bought last month outweighs an industry veteran with opinions.

The message to send — copy it, fill the brackets, send it today

Hi [name] — I'm trying to understand how [people like you / firms like yours] actually handle [the problem — e.g. taking on a new client], and I'd value twenty minutes of your honesty. You came to mind because you recently [bought / switched / dealt with this].

This isn't a sales call — I won't pitch anything, and there's nothing to prepare. I just want to hear how it actually went, including the annoying parts.

Would twenty minutes this week or next suit you? Phone or video, whichever you prefer.

Why it works: it asks for a fixed, small amount of time; it says why them specifically; it disarms the sales suspicion before it forms; and "including the annoying parts" signals you want truth, not applause.

When someone says no — and some will — thank them in one line and ask one question: "Is there someone you know who's dealt with [the problem] recently?" A no plus a referral is a good outcome. No reply after four days gets one short nudge, then you let it go. This is why the exercise asks you to list three to five recruits for two required interviews: you need spares, not courage.

Twenty minutes without leading the witness

Here is the five-question spine, written out for Herzog Physio. Adapt the nouns; keep the shape.

1. "Think of the last time back pain or an injury made you look for help. Walk me through what happened." — opens on a real event, in their words.
2. "What did you try or use first, and how did you find it?" — surfaces the alternatives, including doing nothing.
3. "What was the hardest part of getting treated?" — the pain behind the pain.
4. "What did the whole thing cost you — time, money, worry?" — the stakes, in their currency.
5. "How did you choose where to go, in the end?" — the real decision process, and where you'd need to show up.

Notice what the spine never does: it never mentions your business, never asks "would you", never proposes a feature. Your job during the answers is mostly silence. Three follow-ups cover nearly every situation: "Tell me more about that." — "What do you mean by 'a hassle'?" — and saying nothing for three seconds, which is uncomfortable and reliably produces the best material. A good interviewer is 20% questions, 80% not talking. Take verbatim notes: the exact words, as taught in Lesson 2.3, are the data.

Consent, in one paragraph

Interviews involve a person's data, so GDPR applies — and compliance here is short. Before recording, ask, and offer notes-only:

"I'd like to record this so I don't scribble while you talk — is that OK? The recording stays private, I'll anonymise anything I quote, and I'll delete it once my notes are written up. Happy to just take notes instead if you prefer."

State your purpose, keep recordings and notes access-controlled, write people up as roles ("accountant, 8-person firm"), never names, and delete recordings once synthesised. Your coursework and your Marketing Plan carry no personal data at any point.

That's the whole lesson. The exercise below takes twenty minutes. The real homework is pressing send.

---