Estimation question

How much storage data will YouTube need for the next year?

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

Whether you can build a bottom-up estimate combining upload volume, video length, and encoding assumptions.

Clarify scope: net new storage needed for video uploaded to YouTube over the next year, accounting for multiple resolution copies.
Estimate upload volume using a known reference point, hours of video uploaded per minute, and scale it to a yearly total.
State a bitrate assumption per hour of standard and HD video to convert hours into gigabytes.
Account for YouTube storing multiple encoded resolutions per video, which multiplies the raw storage need, and state that multiplier assumption.
Multiply yearly upload hours by size per hour by the encoding multiplier to reach a total storage estimate.
Present the result as a range in exabytes or petabytes, and flag the encoding multiplier as the most sensitive assumption.

Anchors the estimate to a known reference, hours uploaded per minute, rather than guessing total hours from scratch.
Explicitly accounts for multiple resolution encodings per video, which most naive estimates miss entirely.
States a labeled bitrate assumption, such as roughly 1 GB per hour of standard-definition video, before scaling up.
Flags the encoding multiplier as the single most sensitive assumption driving the final order of magnitude.

Estimating storage for only one video resolution, ignoring that YouTube stores several encoded copies per upload.
Presenting a suspiciously precise final number instead of a labeled, sanity-checked range.

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