Facebook new-user 30-day retention problem in one country: diagnose, fix and measure

You are a growth marketing analyst at Facebook. In one particular country, people who sign up for a new Facebook account are not coming back once their first 30 days are over. You have been asked to work out why this is happening and what to do about it. Answer the four parts below in order. ### Clarifying Questions - Is this a recent change for this country, or have its new users always retained worse than new users in comparable countries? - How exactly is "returning after 30 days" measured: active on day 30, active at any time in the following weeks, or another definition? - Has anything changed recently in how people in this country find and join Facebook, such as a new campaign, partner deal or sign-up flow? - Does the problem cover every platform and type of new user in the country, or is it concentrated in one group? ### Part 1 — Form hypotheses What hypotheses would you put forward to explain why new users in this country stop returning? ```hint Walk the first month Follow one new user in this country from the sign-up screen to day 30 and ask, at each step, what could make them give up. ``` #### What This Part Should Cover - Hypotheses organized so that the main families of causes are all represented - Explanations tied to conditions in this country, not to Facebook in general - Whether the retention number itself can be trusted ### Part 2 — Prove or disprove each hypothesis For each hypothesis, which data and metrics would you examine, and what result would confirm it or rule it out? ```hint Pick the baseline Before choosing a metric, decide what you will compare this country's new users against. ``` #### What This Part Should Cover - Concrete metrics and cuts for each hypothesis, each with a baseline to compare against - The result that would confirm and the result that would rule out each hypothesis - A sensible order for running the checks ### Part 3 — Prioritize and fix How would you decide which confirmed problem to solve first, and what changes would you propose for it? ```hint Size before you rank Estimate how many of the lost users each confirmed cause actually accounts for before ranking them. ``` #### What This Part Should Cover - Explicit criteria for ranking the confirmed causes - Changes that each target a specific confirmed cause - Side effects of the changes on existing users, account integrity or privacy ### Part 4 — Measure success How would you measure whether the change you ship worked? ```hint Do not wait a month blind A 30-day metric takes at least 30 days to read; think about what you could observe sooner that predicts it. ``` #### What This Part Should Cover - A primary success metric, earlier signals, and guardrails - A test design and how long it must run before the result is trusted ### What a Strong Answer Covers - A precise definition of the retention metric and the cohort before any hypothesis is offered - A clear chain from each hypothesis to the evidence, the fix and the measure of success - Reasoning grounded in the affected country rather than in Facebook-wide assumptions - Awareness that changes aimed at new users also reach the existing network around them ### Follow-up Questions - Suppose the drop is concentrated among users who arrived through one new acquisition channel. Is that a product problem or a marketing problem, and what would you do? - If retention improves in your test but sign-ups in the country fall, how would you decide whether the change succeeded? - How would your diagnosis change if the same retention problem appeared in many countries at once?