Back-of-envelope sizing of YouTube's user base

You are asked to work out how many people use YouTube. Nothing can be looked up: build the figure from assumptions you state out loud, walk through the arithmetic, and finish by checking whether the result is believable. ```hint Pin down "user" first Decide what counts as one user and over what time window before you multiply anything. Then ask which groups of people differ most in how likely they are to watch. ``` ### Clarifying Questions - Should a "user" be any person who watches at least one video in the period, a signed-in account, or someone who uploads videos, and is the answer wanted as a monthly count, a daily count, or both? - Is the scope worldwide or a single country? - Do children watching on a parent's device, households watching together on a smart TV, and people who only play videos embedded on other websites count? ### What a Strong Answer Covers - A precise definition of the metric and its time window, agreed before any arithmetic. - A clear equation from population to YouTube users that removes markets where the service is unavailable. - Segmentation by a factor that really changes YouTube usage, such as age, device or connectivity, with a separate reach rate per segment. - A conversion from monthly to daily users through a frequency mix rather than one unexplained ratio. - A sanity check against something independent, and a note on which assumption moves the result most. ### Follow-up Questions - The interviewer now wants signed-in accounts rather than people. What changes in your estimate? - Your monthly figure lands well above a number the interviewer quotes. How do you find the assumption that is off? - How would you estimate how many of these users uploaded at least one video in the last month?