PM Interview: Fermi Estimation Practice

You Don't Need to Be a Math Genius to Nail Fermi Estimation

Here's a secret about estimation interviews: the interviewer already knows the answer. They're not testing your arithmetic. They're testing whether you can think clearly under uncertainty — which is, incidentally, what product managers do every single day.

A PM at DoorDash once told me that estimating "how many deliveries happen in Manhattan on a Friday night" is essentially the same skill as sizing a new market or forecasting adoption for a feature. Both require you to break an unknowable whole into knowable parts.

The Real Reason PMs Get Estimation Questions Wrong

Most candidates fail Fermi questions not because of bad math, but because of bad assumptions. A 2023 McKinsey analysis of structured problem-solving found that 70% of estimation errors come from the framing step — choosing the wrong starting point or missing a critical variable — not from calculation mistakes.

The classic trap: going too granular too fast. If someone asks "how many piano tuners are in Chicago," a weak answer starts counting neighborhoods. A strong answer starts with "how many pianos exist in Chicago, how often each needs tuning, and how many tunings a tuner can do per day." Three numbers, one division, done.

What interviewers really want to see is your comfort with ambiguity. Can you state an assumption, acknowledge it might be off by 2x, and keep moving? That's a PM skill. Candidates who get paralyzed trying to find the "right" assumption are revealing something about how they'd behave in a roadmap planning meeting.

How This Prompt Helps

This prompt generates realistic estimation problems calibrated to your chosen difficulty and domain. More importantly, it evaluates your approach on the dimensions that actually matter: did you scope the problem, pick a logical approach, state reasonable assumptions, get the math right, and sanity-check your answer?

The model answer walkthrough is where the real learning happens. It shows you a clean solution path so you can compare your reasoning structure against a strong benchmark.

When to Reach for This

• You have a PM interview at a company known for analytical rigor (Google, Stripe, Uber) and want to sharpen your estimation skills
• You're comfortable with product sense questions but freeze when numbers come up
• You want to practice domain-specific estimations (fintech TAM calculations feel different from consumer app usage estimates)
• You need to build intuition for "reasonable" numbers in tech (daily active users, conversion rates, revenue per user)
• You're prepping for case study interviews where market sizing is the opening question

What Good Looks Like

A strong output walks you through a complete estimation with explicit assumptions at each step, a clear top-down or bottom-up structure, arithmetic you can follow, and a sanity check against a known reference point. The final answer should be a range, not a single number — because in the real world, precision is less valuable than calibration.

Sources

• How to Crack Product Management Estimation Questions — McKinsey
• Fermi Estimation: A PM's Secret Weapon — Mind the Product