Estimate concurrent YouTube viewership across India at a single moment

Quick Overview

A product estimation exercise asking candidates to size how many people across India are watching YouTube at one moment. It tests scoping by time of day and device, building an equation that turns a user base into simultaneous viewers, labeling round-number assumptions, and sanity-checking the result.

Estimate concurrent YouTube viewership across India at a single moment

Company: Amazon

Role: Product Manager

Category: Product / Decision Making

Difficulty: medium

Interview Round: Other

Estimate how many people in India are watching YouTube at this moment. Talk through how you would pin down what the question counts, the equation you would build, the assumption behind every input, and how you would check that the final figure is believable. ```hint Moment, not day A daily user count is not the number of people watching in a given minute. Think about what links total daily viewing to how many people are watching at once, and how unevenly that viewing spreads across 24 hours. ``` ### Clarifying Questions - What time is "right now" in India: a weekday morning, a working afternoon, or evening prime time? The answer can differ several times over. - Which surfaces count: the phone app, the website, smart TVs, YouTube players embedded on other sites, and short clips as well as long videos? - Are we counting people or screens? A family in front of one living-room TV is several viewers on a single device. - Is a single number enough, or should the answer show how it changes over the day? ### What a Strong Answer Covers - Scoping decisions that materially change the number (time of day, surfaces, people versus screens), each stated as an assumption. - An equation that converts a user base into the number watching simultaneously by accounting for how long each person watches, rather than stopping at daily users or video plays. - Labeled round-number inputs for population, internet access, YouTube reach and daily engagement, with a reason for each. - A time-of-day adjustment and a final range rather than false precision. - A sanity check by a second route. ### Follow-up Questions - A cricket final is streaming tonight on a rival service. Which inputs in your model move, and in which direction? - Which single assumption is your estimate most sensitive to, and what data could you realistically obtain to pin it down? - An ad sales team asks how many ad slots YouTube could fill in India at 9 PM. How would you turn your viewer estimate into an impression estimate?

Overview: A product estimation exercise asking candidates to size how many people across India are watching YouTube at one moment. It tests scoping by time of day and device, building an equation that turns a user base into simultaneous viewers, labeling round-number assumptions, and sanity-checking the result.

Amazon logo
Amazon
Feb 10, 2025
mediumProduct ManagerOtherProduct / Decision Making
0
0

Estimate how many people in India are watching YouTube at this moment. Talk through how you would pin down what the question counts, the equation you would build, the assumption behind every input, and how you would check that the final figure is believable.

Clarifying Questions Guidance

  • What time is "right now" in India: a weekday morning, a working afternoon, or evening prime time? The answer can differ several times over.
  • Which surfaces count: the phone app, the website, smart TVs, YouTube players embedded on other sites, and short clips as well as long videos?
  • Are we counting people or screens? A family in front of one living-room TV is several viewers on a single device.
  • Is a single number enough, or should the answer show how it changes over the day?

What a Strong Answer Covers Guidance

  • Scoping decisions that materially change the number (time of day, surfaces, people versus screens), each stated as an assumption.
  • An equation that converts a user base into the number watching simultaneously by accounting for how long each person watches, rather than stopping at daily users or video plays.
  • Labeled round-number inputs for population, internet access, YouTube reach and daily engagement, with a reason for each.
  • A time-of-day adjustment and a final range rather than false precision.
  • A sanity check by a second route.

Follow-up Questions Guidance

  • A cricket final is streaming tonight on a rival service. Which inputs in your model move, and in which direction?
  • Which single assumption is your estimate most sensitive to, and what data could you realistically obtain to pin it down?
  • An ad sales team asks how many ad slots YouTube could fill in India at 9 PM. How would you turn your viewer estimate into an impression estimate?