I froze for a second because I kept second-guessing my baseline numbers.
Break down the problem by estimating Facebook's total user base, then apply assumptions about daily active usage, peak concurrency, and session duration to derive the number of concurrent users. Use a structured, top-down approach with clear assumptions and sanity checks.
Pro tip: State your assumptions explicitly and round numbers to simplify calculations; interviewers care more about your logical process than the exact number. Also, mention that the estimate should be a range, not a single point, to reflect uncertainty.
Ask if 'people on Facebook' means logged-in users actively using the platform (e.g., scrolling, posting) or just having the app open. Also confirm if we're estimating globally and at what time (e.g., peak vs. average).
Start with world population (~8B), internet penetration (~60%), and Facebook's market share among internet users (~60-70%). This gives roughly 2.9-3.4B registered users.
Assume a DAU/MAU ratio of about 60-70% for Facebook, yielding ~2B daily active users. This accounts for users who log in at least once per day.
Assume the average DAU spends about 30-60 minutes per day on Facebook, spread across peak and off-peak hours. At peak, concurrent users might be 10-20% of DAU, while average concurrency is lower (e.g., 5-10%).
Multiply DAU by the concurrent percentage to get a range (e.g., 2B * 10% = 200M at peak). Sanity-check against known metrics like Facebook's reported 2.9B MAU and 2B DAU, and adjust for regional usage patterns.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.