Estimating Average Reciprocal Follow Connections per TikTok User

Estimate how many mutual follows the average TikTok user has, where a mutual follow is a pair of accounts that follow each other. Walk through your assumptions and check your result. ```hint Split the follow graph Most follows on TikTok go one way, from viewers to creators they enjoy. Separate follows between people who know each other from follows of creators, and estimate how often each kind is returned. ``` ### Clarifying Questions - Do you want the mean, as "average" suggests, or the typical user's count (the median)? On a skewed network the two can differ a lot. - Should the base be all registered accounts, including dormant ones, or only monthly active users? - Is this worldwide, or for one market such as the US? - Do creator and brand accounts count as users, or only personal accounts? ### What a Strong Answer Covers - A precise definition of a mutual follow and of the user base, with awareness that the mean and median diverge - A structure that segments users by how they use the app rather than starting from world population - Separate follow-back assumptions for follows between people who know each other and follows of creators - Clearly labeled round-number assumptions and arithmetic that checks out - A sanity check through a second method or a comparison with friendship-based networks ### Follow-up Questions - How would your estimate differ for users aged 13 to 17 compared with users over 35, and why? - If you had TikTok's follow data, how would you compute this number exactly, and which product decisions would it inform? - Which product changes could raise the number of mutual follows, and why would TikTok want that?