Magic Questions (Statements That End in 'Do You Agree?')
To understand how someone thinks, feed them statements ending in 'is that right?' rather than asking open-ended questions.
- Difficulty
- Moderate
- Time to result
- ~days to results
- Steps
- 4
- Confidence
- 90%
A technique for reverse-engineering another person's mental model — especially people who can't or won't articulate their reasoning. Instead of asking an open-ended 'why' question, you put a specific concrete assertion in front of them and watch what they say yes and no to. Over several such statements you triangulate the underlying model without ever asking them to explain it. Gridley also has her own reports use it on her, which trains their judgment and stops them becoming dependent on her for answers.
Origin
Developed by Hilary Gridley, originally to navigate legal and compliance teams in regulated healthcare spaces who could only answer 'it depends' to open-ended questions.
Core principles
- 01Magic questions are not actually questions — they are statements that end with 'do you agree?' or 'is that right?'
- 02People are often bad at explaining their own reasoning; watching what they affirm or reject reveals more than asking why.
- 03You must go in 'pure of heart' — genuinely open to being wrong — or it becomes coercive and yields bad intel.
- 04Having someone assert a position and then calibrate against your reaction trains their judgment faster than handing them the answer.
- 05It prevents reports from becoming reliant on the manager for every answer.
How to run it
- 1
Convert your question into a concrete statement
Instead of 'what should I do?' or 'why are we doing this?', phrase your best guess as a specific assertion: 'What if the copy said X — would that be okay?'
Pro tip Make the statement concrete and testable, not abstract; specifics force a clear yes/no.
- 2
End with 'do you agree?' or 'is that right?'
Append the calibration prompt so the other person only has to react, not construct an explanation from scratch.
- 3
Note what they affirm and reject, and probe why
Track each yes and no. If they can explain the reason, great; if not, keep offering new statements to triangulate the model.
Pro tip When teaching reports, tell them: 'Rephrase that as — tell me what you think you could have done differently, and then ask me if I agree.'
- 4
Approach with visible curiosity and humility
Signal genuinely that you expect to be wrong and are trying to understand, not to extract a 'yes'.
Watch out If you come in hostile or forceful, or your goal is to get to 'yes', the person gets uncomfortable and gives you bad intel — they'll tell you what you want to hear.
In the wild
When Gridley first worked in a regulated space, legal and compliance teams answered every open-ended 'what's the rule?' with 'it depends.' She flipped her approach: 'What if we did X, Y, and Z — what if this is what the copy said, would that be okay? If no, why? If yes, why?' She teased out the mental model rather than asking them to explain it.
→ She could reliably determine what was and wasn't permissible in ambiguous regulatory areas without a clear written rule.
Common mistakes
Asking open-ended questions of someone whose model you want
Open-ended 'why' questions yield vague or 'it depends' answers and slow calibration; a statement to react to is far faster.
Going in to get a 'yes'
If your goal is agreement rather than understanding, the technique becomes coercive, the person senses it, and you get unreliable answers.
Is it for you?
Best for
Product managers and operators trying to decode a leader's, legal team's, or stakeholder's decision criteria, and managers coaching reports to build independent judgment.
Not ideal for
Relationships where trust is low or the dynamic is adversarial — the person will read the statements as manipulation and mislead you.
From the transcript
“they're not actually questions. their statements and they end with do you agree or is that right?”
“I'm sort of teasing out the mental model rather than asking them to explain it to me”
“you got to go in pure of heart”
From the episode
How to build a team that can “take a punch”: A playbook for building resilient, high-performing teams
Hilary Gridley (Head of Core Product, Whoop)