Skip to content
PMcademy
Template

The user interview guide

A discovery guide built on what people actually did, not what they say they'd do next.

When you reach for it

Before any round of user interviews, especially the first one on a new problem.

  1. 1

    Screener

    Makes sure you're talking to someone who has the actual behaviour you're studying, not someone willing to imagine having it.

    What it asks

    What did this person actually do, recently, that qualifies them, not what they say they would do?

    Filled in

    Screen for: linked a bank account in Sona within the last 30 days, and set at least one budget category. Not: "would you be interested in budgeting features," which qualifies anyone who says yes to a hypothetical.

    How it goes wrong

    Screening on stated interest ("are you someone who cares about budgeting") instead of actual behaviour. Everyone says yes to caring about budgeting. Far fewer have ever set one.

  2. 2

    Opening: the last time it happened

    Anchors the whole interview in a specific, recent, real event instead of a general opinion.

    What it asks

    What's the most recent specific instance of the behaviour you want to understand? Ask them to walk through it.

    Filled in

    Tell me about the last time you set a budget in Sona. Not this week in general, the actual last time. Where were you, what made you open the app right then?

    How it goes wrong

    Asking "how do you usually budget" instead of "tell me about the last time." "Usually" invites a generalised, tidied up answer. "Last time" forces a specific memory with the mess still in it.

  3. 3

    Follow the friction, not the feature

    Keeps the conversation on what actually went wrong or felt hard, rather than drifting into a feature request session.

    What it asks

    At each step they describe, ask what almost made them stop, or what they weren't sure about.

    Filled in

    You said you weren't sure which category to pick for that purchase. What did you do? Did you guess, skip it, or go looking for help? What would have told you the answer faster?

    How it goes wrong

    Letting the conversation drift to "what feature would you want," which produces a wishlist built from whatever they can imagine, usually a worse version of something they've seen elsewhere, instead of the actual friction in their real workflow.

  4. 4

    Never ask about the future

    Removes the invitation to speculate. People are unreliable narrators of their own future behaviour and reliable narrators of what they already did.

    What it asks

    Reframe any "would you use X" question as a question about what they've done in the past that's closest to X.

    Filled in

    Instead of "would you use an automatic savings feature," ask: "have you ever set up an automatic transfer anywhere, in any app or bank, and if so, why that one, and what stopped you doing it in others."

    How it goes wrong

    Asking "would you pay for this" or "would you use this" directly. People are polite and imaginative. They'll say yes to be helpful, and that yes predicts nothing about what they'll actually do.

  5. 5

    Close: what happened right after

    Captures the immediate consequence of the behaviour, which is often more revealing than the behaviour itself.

    What it asks

    What did they do in the minutes or days right after? Did they come back, tell someone, give up?

    Filled in

    After you set that budget, what happened next? Did you check it again? When? What made you check, or what made you forget about it?

    How it goes wrong

    Ending the interview at the moment of the action instead of following through to the consequence, which is where you learn whether the feature actually changed behaviour or just produced a one time action that was immediately forgotten.

Take the whole thing

Paste into your doc
# User interview guide: [research question]

## Screener
Qualifying behaviour in the last 30 days (not stated interest):


## Opening
Tell me about the last time [behaviour happened]. Walk me through it.


## Friction, step by step
At each step: what almost made you stop? What weren't you sure about?

1.
2.
3.

## Past behaviour, not future intent
Instead of "would you use X," ask what they've done closest to X already.


## Close
What happened right after? Did you come back, tell someone, give up?


Duration booked: 45 min. Plan to use: 30 min.
More templates