I started with storage and bandwidth since those feel obvious for a video platform, but then blanked on how to size the compute layer.
Break down YouTube's daily costs into major categories: infrastructure (storage, bandwidth, compute), content acquisition (licensing, creator payouts), and operations (R&D, marketing, overhead). Estimate each component using available public data and reasonable assumptions, then sum them to get a total daily cost. Focus on demonstrating structured thinking and prioritization rather than getting the exact number.
Pro tip: Anchor your estimate by comparing to Google's known financials (e.g., total revenue, costs) and YouTube's estimated share, then validate with a bottom-up approach. This shows you can leverage both top-down and bottom-up thinking, a key PM skill.
Define what 'run YouTube' includes (e.g., infrastructure, content, operations) and state key assumptions like daily active users, average watch time, and storage needs.
Calculate costs for bandwidth (video streaming), storage (video library), and compute (transcoding, recommendations) using estimated data volumes and cloud pricing.
Estimate daily creator payouts (e.g., from ad revenue share) and content licensing costs (e.g., music rights, premium content).
Allocate a portion of Google's R&D, marketing, and general overhead to YouTube based on headcount or revenue share.
Add up all components to get total daily cost, then sanity-check against public estimates (e.g., YouTube's annual costs) and adjust for reasonableness.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.