Cutting Google Photos' storage footprint with infrastructure and product levers

You are a product manager on Google Photos, the service that backs up and stores people's photos and videos. You are asked how you would reduce the total amount of storage the service requires. Explain where the storage goes, which changes you would make to bring it down, and the order in which you would make them. ```hint Break the total into factors Write total storage as a product of a few factors, such as how many items are kept, how large each one is, and how many physical bytes back each logical byte. Then ask which factor is largest and which can shrink without users noticing. ``` ### Clarifying Questions - Is the main aim lowering Google's infrastructure cost, helping users fit more into their quota, or both? - Must files that users uploaded at original quality come back bit-for-bit identical, or is visually lossless re-encoding acceptable with consent? - Do we know how the bytes split between photos and videos, and between recent and rarely viewed items? - Are changes to storage policy, such as how long inactive accounts keep their content, within scope? ### What a Strong Answer Covers - A decomposition of total storage into its drivers, with a reasoned view of which ones dominate (for example video versus photos, originals versus derived copies, redundancy overhead). - Infrastructure levers users never see, with an honest note on which reduce bytes and which only reduce cost per byte. - User-facing levers, such as cleanup suggestions and quality choices, designed so users stay in control of their memories. - Prioritization by bytes saved, user impact, risk and effort, including the tension with storage-plan revenue. - Metrics that prove the savings, and guardrails that protect against data loss and visible quality loss. ### Follow-up Questions - How would you detect near-duplicate photos, and what should the product do when it is unsure two shots are duplicates? - A re-encoding pass would save a lot of space, but a small share of users notice the quality drop. How do you decide whether to ship it? - Storage plans earn money. How do you cut storage without undermining that business?