Ad platforms confuse a lot of PM candidates. The systems are technical, the jargon is dense, and most prep guides skip the topic. If you are interviewing at Google, Meta, Amazon, Reddit, Pinterest, TikTok, or any company with an ad-supported product, you need a clear mental model of how ads actually work.
This post covers the three sides of an ad platform, the mechanics of an ad auction, and the targeting signals that drive revenue. By the end, you will have the vocabulary to handle a product sense question on ads or a strategy question on monetization.
The three-sided market
Every ad platform has three customers: users, advertisers, and publishers. Users see the ads. Advertisers pay to show ads. Publishers own the inventory where ads appear.
On a closed platform like Meta or TikTok, the company is both the platform and the publisher. On an open platform like Google Ads on third-party sites, publishers are independent websites and apps. Either way, the PM job is to balance the needs of all three sides.
Users want relevance and a clean experience. Advertisers want conversions at a price that beats other channels. Publishers want high revenue per impression. These goals create real tradeoffs. More ads per page raises short-term revenue but hurts user retention. Stronger targeting raises advertiser ROI but can feel invasive to users. As Marty Cagan notes, product teams that ignore the business side of these tradeoffs end up shipping features that nobody can monetize (Cagan 47).
For interviews, anchor every answer in this three-sided market. If a question asks how to grow ad revenue, your answer should touch on all three sides, not just one.
How ad auctions work
The core mechanic of every modern ad platform is the auction. When a user loads a page or scrolls a feed, the platform runs an auction in milliseconds to decide which ad to show.
Most platforms use a second-price auction or a variant. In a second-price auction, the highest bidder wins, but they pay the price of the second-highest bid plus a small increment. This design encourages advertisers to bid their true maximum value, since they will not overpay.
The winning ad is not always the one with the highest bid. Platforms rank ads using a formula that combines the bid with a quality score. A simplified version looks like this:
Ad rank = bid amount × predicted click-through rate × ad quality
This means a $2 bid on a highly relevant ad can beat a $5 bid on a junk ad. The platform wins because users get better ads, advertisers with strong creative pay less, and publishers get higher engagement. According to Lenny Rachitsky, this kind of multi-objective ranking is the engine that drives long-term ad revenue at companies like Google and Meta (Rachitsky).
A common interview question: "How would you change the ad auction at Pinterest?" A weak answer changes the bid weight. A strong answer asks what metric is being optimized, looks at the impact on all three sides, and proposes a measurable test.
Targeting signals
Ad targeting is how platforms match an ad to a user. The signals fall into a few buckets.
Demographic targeting uses age, gender, location, and language. This is the oldest form of targeting and the least precise. Behavioral targeting uses what a user has done on the platform, such as pages liked, videos watched, or products viewed. Contextual targeting uses what is on the page right now, such as keywords in the article. Lookalike targeting finds users who resemble an advertiser's existing customers.
The most powerful signals come from first-party data. A user logged into Amazon who just searched for running shoes is a high-value impression for Nike. A logged-out visitor on a news site is much harder to target.
Privacy regulations have reshaped this space. Apple's App Tracking Transparency, Europe's GDPR, and the slow death of third-party cookies have forced platforms to lean harder on first-party data and probabilistic models. PMs who interview in ad tech should be ready to discuss the post-cookie world. A good starting point is the Interactive Advertising Bureau's reporting on identity and measurement (IAB).
Key metrics for ad PMs
You should know these numbers cold.
CPM is cost per thousand impressions. CPC is cost per click. CPA is cost per acquisition. CTR is click-through rate. Conversion rate is the share of clicks that complete the desired action. ROAS is return on ad spend, which is revenue divided by ad cost.
On the platform side, fill rate is the share of ad slots that get filled with a paying ad. eCPM is effective cost per thousand impressions, which lets you compare revenue across different pricing models. Ad load is the share of content that is ads.
In an interview, if you are asked to diagnose a drop in ad revenue, walk through the funnel. Did impressions drop? Did fill rate drop? Did eCPM drop? Each points to a different root cause and a different fix.
How to answer ad platform interview questions
Three patterns come up again and again.
The first is product sense: "Design an ad product for X." Start with the user, then the advertiser, then the publisher. Pick one job to be done for each side. Propose a format that fits the surface. Define the auction rules. Pick two or three success metrics.
The second is strategy: "Should company X enter the ad business?" Look at the asset they own. Do they have logged-in users with strong intent? Do they have inventory? Do they have first-party data? Compare the unit economics of ads against subscriptions or transactions.
The third is analytical: "Revenue is down 10 percent. What do you do?" Use the funnel. Segment by surface, geography, advertiser vertical, and user cohort. Form a hypothesis before pulling more data.
The candidates who do well are the ones who treat ads as a real product with real users on every side. The ones who struggle treat ads as a tax on the user experience. As Teresa Torres reminds us, good product work starts with continuous discovery of what users actually need (Torres 23). That applies to advertisers and publishers too.
Works cited
Cagan, Marty. Inspired: How to Create Tech Products Customers Love. 2nd ed., Wiley, 2017.
Interactive Advertising Bureau. "State of Data 2024: Identity and Addressability." IAB, 2024, www.iab.com.
Rachitsky, Lenny. "How the Best Product Teams Drive Growth." Lenny's Newsletter, www.lennysnewsletter.com.
Torres, Teresa. Continuous Discovery Habits. Product Talk, 2021.