Setting Yelp's ad load: trading ad revenue against user and advertiser experience
Company: Yelp
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
Yelp earns money by showing ads from local businesses next to its listings and reviews. As a product manager at Yelp, how would you work out the right number of ads to show on a Yelp page?
Explain what "right" should mean here, what the answer depends on, how you would test different ad loads, and how you would turn the results into a decision.
```hint Price each extra slot
Ask what one more ad slot earns and what it costs, and who pays that cost: the searcher whose next organic result moves down, or the advertisers whose clicks get spread across more slots. Some of those costs only appear weeks later.
```
### Clarifying Questions
- Which page do we mean: search results, an individual business's page, or the home feed, and in the app, on the mobile web, or on desktop?
- Is the target near-term ad revenue, or long-term value that also counts returning users and the return advertisers get?
- May the number of ads vary by search, category or city, or must each page type have one fixed number?
- What is the ad load today, and is there any signal that it is too high or too low?
### What a Strong Answer Covers
- A definition of "optimal" that weighs Yelp's ad revenue against user trust and engagement and against advertisers' return, over a long horizon
- The factors that move the right ad load: page type, device, search intent, and how many relevant advertisers a market has
- An ad-load experiment with a sound randomization unit and duration, which accounts for users learning to ignore ads and for advertiser budgets shared through the auction
- A primary metric plus Yelp-specific guardrails such as organic result clicks, leads sent to businesses, and returning users
- A concrete rule for choosing the level from the results, and a plan for revisiting it
### Follow-up Questions
- Over a long test, one extra ad slot raises revenue per search, but searches per user drift slowly downward. How do you decide whether to ship it?
- How would your answer change on a business's own page, where the ads may promote its competitors?
- In a small town with only a handful of advertisers, how many ads should a search results page show?
Overview: A product manager interview question about deciding how many ads Yelp should show on a page. It tests whether a candidate can balance ad revenue against user trust and advertiser returns, design an ad-load experiment with sensible guardrails, and turn the results into a clear decision.