Ask About Behavior, Not Intention
Ask what they did last time, not whether they'd use it.
People can't predict what they'll do next, but they remember what they did last week. So ask only about the past, never the future.
What you'll run into:
- They nod through the whole interview, then don't use what you build
- The survey results look great, and the conversion numbers say otherwise
- You finish the round of interviews with a notebook full of adjectives and not one concrete event
It's not that users are being evasive. When answering a hypothetical question, people picture an idealized version of themselves—one with plenty of time, no inertia, and a willingness to learn new things. The real person will fall back on the familiar old method next week.
There's another layer: the person you're asking wants you to feel good. You took the time to come, so they can't bring themselves to say the thing is useless. Asking about intention is basically inviting a polite answer. This is the starting point of the whole book The Mom Test: don't ask people whether they'd buy—watch how they actually handled the problem last time. mom-test
Even cheaper than asking is watching: have them perform the task right now. When the interview setting allows, ask them to open the tool they actually use and walk you through it—the demo itself is the answer, no question needed.
How to rephrase
The rule is simple: replace every "would you / do you want to / do you need" with "last time / at the time / how did you."
| Don't ask | Instead ask |
|---|---|
| Would you use this feature? | When did you last run into this situation, and how did you handle it? |
| Would you pay for it? | Are you already spending money on something like this? How much? |
| What do you think of this design? | Try to complete this task now. I'll stay quiet; talk me through it as you go. |
| How often would you use it? | When was the last time? And the one before that? |
| What features would you like added? | When did this last make you frustrated, and at which step? |
The second-to-last row is especially useful. Asking "how often do you use it" gets you an estimate; asking "when were the last two times" gets you a real sample of frequency.
Which signals are hardest
Rank them by credibility and you'll know which direction to probe.
- They've already paid. Bought a tool, hired someone, subscribed to a service. The hardest evidence there is.
- They've already cobbled something together. Built their own spreadsheet, wrote a script, redo it by hand every week.
- They can recount one specific incident. With a time, a scene, and details. Credible—but check whether it's an outlier.
- They say the feature is important. An opinion with no facts behind it is basically not decision material.
- They say they'll use it or buy it in the future. Pure prediction. Nice to hear, nothing more.
The goal of an interview is to dig toward the top two tiers. Hearing "right now I manage it with my own spreadsheet" is worth more than ten "it's important"s.
How to use these signals. After the interviews, classify every note into the five tiers. If a tier stays empty—say, nobody has ever paid—you haven't found the problem worth solving yet, so don't rush into building. And if two or three people are already paying for or rigging something around the same problem, that's stronger evidence than any survey number.