Back to library
SurveyMonkeySurveyMonkey

Fermi Estimation Under Pressure

Growth & Marketing

What it tests

Structured thinking with incomplete data and comfort making assumptions out loud

Format

  1. 1Candidate receives an ambiguous estimation prompt with no historical data — e.g. 'Estimate weekly toll revenue for the Golden Gate Bridge given only a blueprint'
  2. 2They have 15 minutes to work through the problem live, talking aloud the entire time
  3. 3Interviewer does not help but may ask 'what else would you need?' to probe assumptions
  4. 4Candidate must land on a specific number with a confidence range, not just a framework

What to look for

  • Do they decompose the problem into discrete, logical sub-estimates — or freeze and hand-wave?
  • Are their assumptions explicitly stated and reasonable, or buried and sloppy?
  • Do they adjust on the fly when a sub-estimate feels off, or commit rigidly?
  • Is the final number defensible, even if wrong — can they explain exactly how they got there?

Adaptation guide

Swap the prompt for something relevant to your product: 'Estimate how many users complete onboarding in the first 7 days' or 'How many support tickets does a 100K MAU SaaS company get per week?'

Full description

Format:

  1. Candidate receives an ambiguous estimation prompt with no historical data — e.g. "Estimate weekly toll revenue for the Golden Gate Bridge given only a blueprint"
  2. They have 15 minutes to work through the problem live, talking aloud the entire time
  3. Interviewer does not help but may ask "what else would you need?" to probe assumptions
  4. Candidate must land on a specific number with a confidence range, not just a framework

Time: 15 minutes

What to look for:

  • Do they decompose the problem into discrete, logical sub-estimates — or freeze and hand-wave?
  • Are their assumptions explicitly stated and reasonable, or buried and sloppy?
  • Do they adjust on the fly when a sub-estimate feels off, or commit rigidly?
  • Is the final number defensible, even if wrong — can they explain exactly how they got there?

Adaptation: Swap the prompt for something relevant to your product: "Estimate how many users complete onboarding in the first 7 days" or "How many support tickets does a 100K MAU SaaS company get per week?"