I started with storage and bandwidth because those felt obvious, then realized I was probably underselling the compute side.
Break down YouTube's annual costs into major cost drivers: content delivery (bandwidth/CDN), storage, content acquisition/licensing, and operations (R&D, sales, marketing). Estimate each component using known metrics (e.g., daily watch hours, average bitrate, storage per video) and industry benchmarks, then sum to a total. Clearly state assumptions and sanity-check the final number against Google's overall financials.
Pro tip: Anchor your estimate to a known figure like Google's total cost of revenues or YouTube's estimated revenue, and show how your bottom-up estimate aligns with that top-down view. This demonstrates business acumen and avoids getting lost in the weeds.
Define what 'running YouTube' includes (e.g., infrastructure, content costs, personnel) and state key assumptions like time period (annual) and global scale. Ask clarifying questions if needed.
List the primary cost drivers: bandwidth/CDN, storage, content acquisition (licensing, creator payments), and operating expenses (R&D, sales, marketing, admin).
Use known metrics (e.g., 1 billion hours watched daily, average bitrate ~5 Mbps) to calculate bandwidth and storage costs. For content, estimate based on YouTube's revenue share to creators and licensing deals. For operations, use headcount and industry benchmarks.
Add up the components to get a total annual cost. Compare with Google's disclosed costs (e.g., traffic acquisition costs, other cost of revenues) to ensure plausibility.
Present the final estimate with a range, highlighting key uncertainties and the most impactful assumptions. Tie back to product strategy implications if relevant.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.