Estimate What It Costs Facebook to Acquire and Serve One More User

Estimate how much it costs Facebook to add one additional user. Decide which costs belong in the figure and why, build the estimate from clearly labeled assumptions, and check that the result is plausible. ```hint Once versus every year Some costs occur once, when the person is won over; others recur for as long as they stay active. Also ask which of Facebook's costs actually change when a single person joins. ``` ### Clarifying Questions - Does "cost" mean what Facebook spends to win the user (advertising, preinstall and carrier deals), the ongoing cost of serving them (storage, computing, content delivery, moderation), or the total over a period such as the first year? - Is the additional user a typical new sign-up, and from which region? A heavy video watcher on a fast network costs far more to serve than a light user on a low-end phone. - Should fixed costs that do not change for one user, such as engineering salaries and existing data centers, be excluded, or should long-run capacity spending be spread across users? - Is this the Facebook app alone, or should messaging and the company's other apps be included? ### What a Strong Answer Covers - A precise definition of the cost being estimated, with one-time acquisition cost separated from the recurring cost to serve. - A clear line between marginal and fixed costs, with organic, network-driven sign-ups treated differently from paid ones. - A component-by-component equation in which every input is a labeled assumption with round numbers. - Correct arithmetic and a sanity check against an independent reference point such as revenue per user. - How the figure varies across user segments and regions, and which decision it would inform. ### Follow-up Questions - How would the estimate change for a user who mostly watches short videos compared with one who mostly reads text posts and sends messages? - If heavier machine-learning ranking made each user ten times more expensive to serve, in which markets would new users stop paying for themselves? - How would you measure the true acquisition cost when most new users join because friends invited them? - Suppose the interviewer narrows the question to acquisition only: how much does Meta spend to win one new Facebook user, with the ongoing cost of serving that user out of scope? How would you estimate cost per new user as yearly acquisition spend divided by yearly new users?