Estimation question

Estimate hardware costs per Google Photos user.

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Structured Fermi estimation with clearly stated, reasonable assumptions and a sanity check against a known anchor.

State the goal: estimate storage hardware cost per active user, driven mainly by average storage used per user.
Assume average storage per active user, for example 5GB across a mix of light and heavy free-tier users, marked explicitly as an assumption.
Estimate raw storage cost per GB using known cloud storage economics, roughly $0.02 to $0.03 per GB per year at large infrastructure scale, an assumption.
Multiply: 5GB times about $0.025 per GB per year gives roughly $0.12 to $0.15 per user per year in raw storage cost.
Add redundancy and replication overhead, typically 2 to 3x for durability, roughly tripling the estimate to around $0.35 to $0.45 per user per year.
Sanity-check against known paid storage pricing, like 100GB for about $20 a year, confirming the estimate is a small fraction of what's charged, consistent with cloud margins.

States every assumption explicitly and marks it as such, the key expectation for estimation questions.
Uses a real anchor, consumer storage pricing, to sanity-check the bottom-up estimate, a strong estimation technique.
Accounts for replication overhead, showing real infrastructure awareness, not just naive storage-only math.

Giving a number with no stated assumptions or reasoning shown.
Ignoring redundancy and replication, which meaningfully changes real infrastructure cost.

Chapters of the AI PM course that teach what this question tests.