I started with daily active users and average session length, which felt right, but then I got fuzzy on bitrates and just picked a number that sounded reasonable.
Break the problem into a demand-side estimate: calculate daily active users (DAU) on YouTube, average videos watched per user, average video length, and average bitrate. Multiply these to get total data streamed, then sanity check against known internet traffic or YouTube's storage/bandwidth figures.
Pro tip: State your assumptions clearly and round numbers to make mental math easy; interviewers care more about your structured thinking than the exact final number.
Start with world population (~8B), internet penetration (~60%), and YouTube's share of internet users (~1/3). This gives roughly 1.6B DAU, but round to 2B for simplicity.
Consider that some users watch many short videos while others watch few. Assume an average of 5 videos per user per day, acknowledging heavy users on mobile.
Assume average video length of 5 minutes (300 seconds) and average streaming bitrate of 5 Mbps (0.625 MB/s) to account for various resolutions.
Multiply DAU × videos per user × video length × bitrate. For 2B users × 5 videos × 300s × 0.625 MB/s = 1.875 × 10^12 MB = 1.875 exabytes per day.
Compare to known figures: global internet traffic is ~3-4 exabytes per day (2023), and YouTube accounts for ~15-20% of that, so 0.5-0.8 EB/day. Our estimate is higher, so adjust assumptions (e.g., lower DAU or bitrate) to align with reality.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.