Contextual segments: what a buyer should know before using one
Part of Contextual targeting
The short answer
A contextual segment is a prebuilt, named group of pages that a buying platform or data partner has classified as being about one subject, such as running or home cooking. A buyer adds it to a line item instead of listing sites. Its pages change as content is recrawled, so test what it reaches before relying on it.
Key takeaways
- A contextual segment groups pages by their content, not people by their behavior, and its membership is set by whoever classified the pages.
- The questions that matter are how pages were classified, which taxonomy and version the segment uses, which languages it reads and how often it is refreshed.
- A small test flight, a placement report and a read of sample pages show what a segment actually reached.
- A custom site list is the better choice when the client needs to approve every site, buy through a deal or cover niche publishers.
- Segment membership moves as pages are recrawled and reclassified, so a test result describes one point in time.
What is a contextual segment in a buying platform?
Definition
- Contextual segment
- A named, targetable group of web pages, apps or videos that share a subject, built by classifying content. The buyer selects the segment; the classifier decides which pages belong to it at bid time.
Buying platforms offer contextual segments in a few shapes. Display & Video 360's category targeting (opens in a new tab) is the platform's own version: the help page says it targets ads based on market verticals and the content of webpages, and that keyword and category targeting is based on a page's content. Data partners sell their own segments inside the same platforms, with names that describe a subject or an interest. Some agencies and brands also commission custom segments built from a list of words or a written description.
| Segment shape | Who decides membership | What the buyer can inspect |
|---|---|---|
| Platform category | The buying platform's own classifier | The category name and its place in the platform's hierarchy |
| Data partner segment | The partner's classifier, licensed through the platform | A name, a description and sometimes a sample of URLs on request |
| Custom segment from keywords or a description | The partner or platform, from the inputs the buyer supplied | The inputs, plus whatever reporting the partner shares |
| Site list or channel | The buyer, site by site | Every domain or app in the list |
What should a buyer ask about how a segment was built?
A segment name is a promise, not evidence. Two segments called running can hold different pages because their classifiers read different signals, apply different thresholds and refresh on different schedules. These questions turn the name into something a planner can judge.
| Question | Why it matters | An answer worth accepting |
|---|---|---|
| Is a page or a whole site classified? | A site level label can carry one section's topic across the whole domain | Page level, or site level with a stated sampling rule |
| Which signals does the classifier read? | Titles and URLs alone can mislead on pages with little text | The main article text, with menus and comments set aside |
| Which taxonomy and which version? | Category IDs differ between taxonomy versions | A named taxonomy and version, with a mapping if it changed |
| Which languages are supported? | Unsupported languages can fall through classification | A list of languages, including Canadian French where the plan needs it |
| Does a parent segment include its children? | Selecting a broad topic can add its subtopics | A clear yes or no, with the child list |
| How often are pages recrawled? | Content changes after classification | A described refresh practice, not just a promise of freshness |
| Can we see sample URLs? | Names hide what the classifier actually matched | A dated sample of pages the segment held |
Version matters more than it sounds. The IAB Tech Lab Taxonomies repository (opens in a new tab) describes the Content Taxonomy as a common language for describing the aboutness of a webpage, application or video, and says Content Taxonomy 3.0 may not be used alongside earlier versions (Content Taxonomy 1.0 through 2.2). A segment mapped to an older version can carry IDs that do not line up with a newer plan. Our guide to the IAB Content Taxonomy covers the version history.
Parent and child behavior differs by platform, so ask each time. In Display & Video 360, the category help page says that targeting a parent category also automatically targets all of the subcategories underneath it.
How can you test what a segment actually reaches?
The reliable test is to buy a little and read what came back. A segment that sounds right on paper can deliver mostly to a handful of large general sites, or to pages that mention the subject in passing.
A five step reach test for one segment
- Run the segment alone on a small line item. Keep other content targeting off, so the placement report reflects the segment and nothing else.
- Pull the served placement report after a few days. Google's Where ads showed (opens in a new tab) help warns that the report is not an exhaustive list of placements and that placements appear only above minimum thresholds of impressions or interactions, so treat it as a sample.
- Sort domains by impressions and read the top of the list. Note how much delivery sits on a few domains and whether those domains are specialists or general portals.
- Open real pages, not just domains. Read several served pages per top domain and mark each as on topic, loosely related or off topic.
- Write the result down with the date. Record the segment name, the date, the share of read pages that fit and the domains that surprised you.
Display & Video 360 offers one more view. Its app, channel and website targeting (opens in a new tab) help explains that a line item's Apps & URLs targeting can download a composite list of every app, URL, collection and channel included. The same page notes that targeting a website can also reach the publisher's app inventory unless environment targeting limits it, which can surprise a buyer reading the report.
When is a custom site list better than a segment?
A segment trades control for convenience. Display & Video 360 describes channels (opens in a new tab) as custom groups of related websites or apps, usable as a preset, targetable list of brand-appropriate publishers or as an excludable list. A site list like that is fixed until someone edits it, which is its strength and its cost.
| Situation | Contextual segment | Custom site list |
|---|---|---|
| The client must approve every site | Hard, because membership changes after approval | Fits, every site is named and versioned |
| The buy runs through a deal | Harder, a deal is agreed with named sellers | Fits, a deal is set up around chosen domains |
| The topic is broad and well covered | Fits, with little setup and wide reach | Workable, but a long list takes time to build |
| The topic is niche or regional | Risky, a broad label can miss specialists | Fits, specialists can be found and added by hand |
| Breaking news changes the content | Reclassification takes time to catch up | The list stays fixed, so suitability needs its own check |
The two can work together. A segment can run as a discovery line item while the site list carries the core budget, and the segment's best domains become candidates for the next version of the list. Our guide to inclusion lists covers how such a list is kept and approved.
How often do contextual segments change?
Continuously, and there is no universal schedule. Display & Video 360's category help says pages are frequently crawled and reclassified to reflect any content changes, and that any recent events or news may take several days to be targetable using category targeting. Its sensitive categories (opens in a new tab) page adds that impressions are sometimes served near unsafe content despite pre-bid blocking, when the platform receives inaccurate URLs, the site is in an unsupported language or content changes between the bid and the ad being served.
Edits to the buyer's own setup also take time. The app, channel and website targeting page says a few hours may pass before Display & Video 360 updates a line item's bidding after a targeting change. A segment tested in March is therefore a different object in June. Repeat the reach test when a flight restarts, when the segment's provider announces a taxonomy change, or when news in the segment's subject area shifts sharply.
What does a segment test look like on a real brief?
Fictional example
A yoga mat and home practice brand in Oregon
A fictional maker of yoga mats and home practice equipment plans English display ads in Oregon, around Portland, Eugene and Bend. The agency's trading desk proposes a prebuilt fitness and wellness segment for reach and a short site list of yoga, mobility and home workout publishers for the core of the plan.
Before launch, the planner asks the segment's provider which taxonomy version it maps to, whether it reads page text or only URLs, and whether its wellness parent includes a diet and supplements child. The answers show that it does, so the planner selects only the fitness children.
A small test line item runs the segment alone for a week. Reading the served pages shows strong fit on two running publishers and a stretching blog, and weak fit on a city events guide whose listings mentioned studio classes in passing. The running and stretching sites are added to the next version of the site list, the events guide is noted, and the test date is recorded beside the segment name.
How does PlacementGraph help when a segment needs a site list beside it?
Frequently asked questions
Do contextual segments work on YouTube?
Treat them with care there. Google Ads' contextual targeting (opens in a new tab) help says that on YouTube, keyword and topic targeting act as a hint while placement targeting is fully supported. A segment that leans on topics may therefore reach YouTube more loosely than a channel list. Our guide to contextual targeting on YouTube covers the signals YouTube offers.
Are contextual segment names standardized across platforms?
No. A platform category, a data partner segment and an IAB Tech Lab category can share a name and hold different pages. The IAB Content Taxonomy gives a common vocabulary, but each provider decides how pages are assigned. Compare segments by their sample pages and their stated taxonomy, never by the label alone.
Can a contextual segment and a site list run on the same line item?
Check how the platform combines them before trying. Google Ads' placement targeting (opens in a new tab) help says ads target any of the selected content in an ad group, matching any of the topics, placements or keywords in an ad group, so adding a topic beside a placement list widens reach beyond the list. Keeping the segment and the list on separate line items makes each one's delivery readable.
Should a segment's test results be shared with the client?
Yes, in plain terms. Share the segment name, the test dates, the domains that took the most delivery and the share of read pages that fit. A client who sees what a segment reached can decide whether its convenience is worth the loss of site by site approval.
Sources
- Category targeting (opens in a new tab)Display & Video 360 Help
- Sensitive categories in Display & Video 360 brand safety targeting (opens in a new tab)Display & Video 360 Help
- App, channel, and website targeting (opens in a new tab)Display & Video 360 Help
- About channels (opens in a new tab)Display & Video 360 Help
- About the Where ads showed report (opens in a new tab)Google Ads Help
- About contextual targeting (opens in a new tab)Google Ads Help
- About placement targeting (opens in a new tab)Google Ads Help
- Taxonomies repository (opens in a new tab)IAB Tech Lab (GitHub)
- Policy position on online behavioural advertising (opens in a new tab)Office of the Privacy Commissioner of Canada, modified August 11, 2025
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.