Explain the Main Online Ad Targeting Methods and When to Use Each

You are asked, as a product manager candidate, to walk through the kinds of online ad targeting you know. For each method, explain what signal it uses to decide who sees an ad or where the ad appears, where that data comes from, and when the method is a good or poor choice for an advertiser. ```hint Sort by what the method keys on Before listing methods, ask what each one actually matches against: the content being viewed, the person viewing it, or the circumstances of the view such as place, device and time. That structure makes the trade-offs between methods easier to compare. ``` ### Clarifying Questions - Should the answer span search, display, video, social and in-app ads, or concentrate on one channel? - Am I answering from the advertiser's side (choosing whom to reach) or the publisher's side (packaging inventory that advertisers can target)? - Should the answer reflect current privacy limits, such as browsers restricting third-party cookies and mobile platforms requiring consent before cross-app tracking? ### What a Strong Answer Covers - A structured taxonomy that separates targeting methods from the identifiers and data sources that power them. - For each method, the signal it uses, whether the data is first-party, third-party or inferred, and how it is matched when an ad is served. - Trade-offs among precision, reach, cost and privacy, tied to campaign goals such as awareness versus conversion. - How privacy changes and regulation affect which methods remain viable, including limits on sensitive categories and protected groups. - How to check that a targeting choice adds value beyond what a broader campaign would have delivered anyway. ### Follow-up Questions - If third-party cookies disappeared from most browsers, which methods would suffer most and what would you rely on instead? - How would you test whether retargeting produces purchases that would have happened without the ads? - As a publisher, how would you let advertisers reach valuable audiences without handing over user-level data?