Are you doing customer interviews wrong?

Are you making these customer interview mistakes?

Interviewing customers is both an art and a skill. Kinda like cooking. 🥘

You can pick up basics from Samin Nosrat in Salt, Fat, Acid, Heat. You can flip on Food Network and learn knife skills so you don’t lop off a finger. 🔪

But you learn a lot of the nuance in cooking by getting your hands dirty. You perfectly salt the chicken and remember that amount for next time. You simmer cinnamon so long it goes from fall-scented glory to bitter bark (ahem, me on Sunday) and note never do that again. 😬

In other words, to consistently make good meals, you need both taught skills and hands-on experience.

The catch: there’s more at stake than ruined dinner #

You need skills and experience to consistently run good customer interviews, too. But unlike cooking, messing up interviews doesn’t just cost you time, burnt veggies, and a bit of embarrassment (“I hope you like Dominos…” 🍕).

When you interview, what’s on the line isn’t dinner—it’s the future of your product.

The qualitative data you gather from interviews informs your “what next?” product decisions. If you routinely gather bad data (because you introduced bias or asked poor questions) you risk bad product decisions. “Garbage in, garbage out,” as they say.

That’s why it’s important to know a few interview dos and don’ts before you start chatting.

Yes, you’ll need practice to nail the nuances. But as every good coach knows, it isn’t simply practice that makes perfect. You can kick a soccer ball for 5 hours/day. But if you’re slamming it with your toe every damn time, you won’t score beautiful goals...even after 10,000 hours.

It’s right practice (strike with the laces, follow-through with the swing) that makes for perfect upper-90 shots. ⚽️

Here are tips you need to apply right practice to your customer interviews.

7 things NOT to do in customer interviews 🙅 #

1. Pick people at random 💬 #

All feedback is not equal. If you’re creating a houseplant app, feedback from someone with a wall of pothos will be more valuable than feedback from your brother, who’d only buy a houseplant if it grew video games. One might use your product; the other certainly won’t.

  • You’re doing it wrong if: you interview anyone and everyone.
  • What to do instead: Pre-qualify conversants. Use surveys, an initial conversation, or other methods to confirm the person is in your target market.

2. Lead the witness ➡️ #

In a recent interview, I made a big mistake. I asked: “What did you enjoy about x product?” The woman on the other end paused and said, “Actually, I wouldn’t say I enjoyed it...” Whoops. My biased question had led the witness. This is a major don’t.

  • You’re doing it wrong if: your question suggests a specific answer.
  • What to do instead: Ask open-ended questions. A better version of my biased question would’ve been, “Tell me about your experience using x?”

3. Dominate the conversation 🙊 #

Someone once told me he writes, “SHUT UP AND LISTEN” on a blank sheet of paper when he interviews. It reminds him that he’s there to listen not talk.

  • You’re doing it wrong if: if you’re speaking more than you’re listening. Repeat after me: This. Is. Not. A. Pitch.
  • What to do instead: Stop yammering. Let silences prompt the other person to add more detail.

4. Speculate about the future 🔮 #

Humans are terrible at predicting what they’ll do or want in the future. You, me, your potential customers...all of us suck at that. Meaning, the data you get from “what would you pay?” or “would you use something like…” is very unreliable.

  • You’re doing it wrong if: you’re asking the other person to speculate on futuristic scenarios.
  • What to do instead: Focus on what they’ve done or are doing. Ask lots of why questions about real and specific situations.

Via User Interviews

5. Furiously take notes ✍️ #

If you’re tearing through notepads faster than Ansel Elgort tore through Atlanta in Baby Driver, I guarantee you’re missing the good stuff.

You may need to take notes sometimes (e.g. an impromptu meeting). But avoid this whenever possible.

  • You’re doing it wrong if: you’re multitasking.
  • What to do instead: record and transcribe the conversation so you can listen intently.

Useful tools for recording:

Useful tools for transcribing:

6. Cherry-pick anecdotes 🍒 #

Like human metal detectors, we’re wired to hunt down data that backs our hypotheses...and ignore the rest. When you do this in your interviews, you cherry-pick the anecdotes you like and miss crucial feedback.

  • You’re doing it wrong if: you highlight only the feedback you expected or hoped for.
  • What to do instead: Re-read your transcripts. Pay extra attention to stuff that’s unexpected, sour, or seems to contradict your hypothesis.

7. Be stiff and awkward 💼 #

I’ve used the term “interviews” so far because it makes sense to most founders. But really, “interview” isn’t a great way to frame these conversations. As Crazy Egg points out, “Your interview should not feel like an interview at all. It should feel like a conversation.”

  • You’re doing it wrong if...the other person feels like they’re applying for a job.
  • What to do instead: The last conversation you enjoyed probably wasn’t stiff or formal. It was comfortable and easy—like chatting with a friend. Bring that demeanor to your customer interviews.

Right practice takes time and effort—keep at it 👏 #

We know you’ve heard “talk to your customers”—from us, and other founders. But we know, too, the how isn’t easy or clear.

If the way forward seems tricky, you’re in good company. In The Mom Test, Rob Fitzpatrick admits, “The advice that you ‘should talk to customers’ is well-intentioned, but ultimately a bit unhelpful. It’s like the popular kid advising his nerdy friend to ‘just be cooler.’ You still have to know how to actually do it.”

Truthfully, we’re still learning how to do interviews ourselves. We hope this newsletter (which outlines the how we’ve learned so far) helps you keep learning, too.

P.S. If you’re looking for other practical interview tips, check out: