Estimation question

Estimate the total number of videos on YouTube.

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 estimation: breaking a huge unknown number into defensible, stated assumptions.

Anchor on YouTube's known scale, roughly 2.5 billion logged in monthly users, and estimate uploader share.
Assume a small fraction, say 5 percent, of users have ever uploaded at least one video.
Estimate average videos per uploader over the platform's 19 year life, say 5 videos per active uploader.
Multiply: 2.5 billion times 5 percent times 5 videos gives roughly 625 million, then round for scale.
Sanity check against known data points, such as reports of 500 hours uploaded per minute, to cross validate order of magnitude.
State the final estimate as a range, for example 500 million to 1 billion videos, not a false precise number.

States every assumption explicitly and labels them as assumptions, for example '5 percent uploader rate, an assumption'.
Cross checks the bottom up estimate against a known public data point (hours uploaded per minute) for sanity.
Gives a final answer as a range rather than a single overconfident number.
Shows the math cleanly so the interviewer can follow and challenge individual assumptions.

Giving a single precise number with no visible reasoning, which signals guessing rather than estimation skill.
Skipping a sanity check against any known reference point.
Ignoring video deletions and platform policy removals that shrink the live catalog over time.

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