Topics Targeting

Topics targeting works off Google's own classification of the page, video, or channel, not off keywords you supply or signals about the person viewing it. Google's crawlers and models tag inventory across the Display Network and YouTube against a hierarchical topic tree, and picking a topic in Google Ads makes your ad eligible on anything tagged under that node, at whatever depth you selected. Choosing a parent-level topic like Autos & Vehicles pulls in everything nested beneath it, including sub-topics you may not have intended, like Motorcycles or Trucks, because the parent selection inherits the full branch below it rather than acting as a looser version of a specific child topic.

Topics targeting lives inside standard Display campaigns and standard Video campaigns' content targeting section; it is not exposed in Performance Max or Demand Gen, both of which rely on Google's own automated audience signals and placement selection rather than a manual topics list, so an advertiser trying to replicate a Topics setup inside either of those campaign types will not find the option. Topics itself always restricts delivery; there is no observation-only version of a topic. What does carry a Targeting-versus-Observation choice is any audience segment or demographic layer stacked alongside the topic - set that layer to Targeting and delivery narrows to the intersection of the topic and the audience, set it to Observation and the audience is only reported on, with the topic alone still deciding where the ad is eligible to show, and mixing the two up is a common source of confusion about why adding an audience layer either shrank a campaign or seemingly did nothing at all. Placement exclusions and Google's content exclusion controls, digital content labels, sensitive categories, embedded video, parked domains, sit alongside Topics as the tools that actually manage inventory quality, since topic classification by itself says nothing about brand safety.

Topics targeting is most useful for prospecting and awareness plays at scale on Display or YouTube, entering a new vertical with no seed audience or pixel history to build from, or layering broad contextual reach underneath a tighter audience signal in Targeting mode. It matters less, and is often the wrong tool, for tight direct-response campaigns running on thin budgets with a firm CPA target, since a topic node can span an enormous and uneven pool of inventory, and Performance Max or Demand Gen will typically produce more efficient delivery for that kind of goal without the topic tree in the picture at all.

The recurring mistake is selecting a topic at the wrong depth relative to the intended reach: picking a broad parent node expecting a modest, closely related audience and instead getting flooded with tangential inventory the account never explicitly agreed to, since the parent's full branch is included by default. A second mistake is treating a chosen topic as inherently brand-safe because it sounds narrow or professional; topic classification is a content-category label, not a quality or safety filter, so a topic like Business & Industrial can still include low-quality inventory that needs to be trimmed separately through placement exclusions. A third mistake is confusing Topics with content keyword targeting: keywords match on the words present on a page, while topics match Google's whole-page category label, so a page can be classified under a topic without your literal target words appearing on it anywhere, and a page containing your keywords may not carry the topic label you expected at all.

In reporting, pull the Placements report segmented by the Topics you have running to see which specific URLs, channels, or apps within that topic actually served impressions; this is the fastest way to catch a parent-level topic pulling in inventory nobody wanted before it burns meaningful budget. Expect click-through rate on topic-targeted Display and Video inventory to run lower than keyword- or audience-based targeting, since topic reach is inherently broader and less intent-driven; a low click-through rate alone is not a red flag on a campaign built for awareness or view-through goals, and should be read alongside view rate and reach rather than as a standalone verdict.