Topics API

The Topics API is a browser-level feature, part of Google's Privacy Sandbox initiative in Chrome, that assigns a small number of broad interest categories, such as "Fitness" or "Travel & Transportation," to a user based on their recent browsing history. Sites can request these topics through the browser and use them for interest-based ad targeting, without planting a cross-site identifier the way a third-party cookie does.

How it works#

Each week, Chrome looks at the sites a person visited and maps a subset of them to entries in a fixed taxonomy of topics. The browser then picks a handful of the most represented topics for that period and stores them locally, on-device. When a participating site calls the API, the browser hands back one topic per recent period, not a running profile and not a device-level ID. There's no server-side database mapping a person to their full topic history; the calculation and storage both happen in the browser itself.

A few design choices matter for how this plays out in practice. Topics rotate on a recurring schedule rather than accumulating indefinitely, so targeting reflects recent interest rather than a permanent label. The taxonomy itself is curated, publicly documented, and excludes sensitive categories (health conditions, sexual orientation, and similar). And a site only sees topics for users who've visited topic-eligible sites recently, so coverage is uneven, especially early in adoption.

Why it exists#

Topics API is Google's answer to a problem every native and programmatic buyer already feels: third-party cookies are unreliable in Safari and Firefox, restricted by policy, and increasingly blocked at the browser level even where they still technically work. Interest-based targeting used to lean on a cookie ID that followed a user across sites; Topics API tries to preserve some of that targeting value using coarse, on-device categories instead of a persistent cross-site key. It sits alongside other Privacy Sandbox pieces (like attribution reporting) as one of several browser-native alternatives publishers and ad networks can adopt as cookieless targeting options mature. It is not a drop-in replacement for the granularity a tracking cookie offered, and adoption across demand-side platforms and exchanges has been gradual rather than universal.

What it means for native advertising#

Native networks like Taboola, Outbrain, and MGID already lean heavily on contextual targeting, matching an ad to the page a reader is on rather than to a tracked identity, so the shift toward Topics-style signals is less disruptive for native than it is for identity-heavy display and social buying. Native buyers who've built campaigns around first-party data and page context, rather than a purchased audience segment, are better positioned as cookie-based real-time bidding signals get thinner across the open web, and pairing contextual signals with solid geo targeting recovers much of the precision a lost cookie would otherwise have provided.

If you're trying to understand how a specific competitor is actually targeting, rather than guessing from a spec sheet, watching the ads they run is more reliable than parsing a privacy whitepaper. OpenAdLibrary's ad intelligence index tracks which verticals and geos each native network's live creatives are concentrated in, so you can see targeting patterns play out in the wild instead of inferring them from documentation.