Contextual Targeting with the IAB Content Taxonomy

Contextual targeting places an ad next to content on the same subject, so the page the reader is on decides where the ad appears, not the reader's history. To choose topics and publishers well, write down what your ad is about in one phrase, name that subject and its neighbours with a shared vocabulary such as the IAB Tech Lab Content Taxonomy, and then compare those names with what publishers actually run on their pages. If you run a small shop, manage an online store, or buy placements for clients as an agency, the steps are the same. An invitation printing studio serves as the running example.

Reading the page instead of the person

Targeting answers two questions: who sees the ad, and where does it appear. Audience-based methods start from the first question. Contextual methods start from the second. A reader looking at a venue checklist is already thinking about the wedding, so an invitation printing studio's ad belongs to the moment instead of interrupting it. That is a practical expectation, not a promise that the ad will perform better.

What changes in your planning

With audience lists you ask, "Which people do I want to reach?" With context you ask, "Which subjects are my buyers reading when a decision is close?" The second question suits a small business well, because the owner can answer it from customer conversations. Example scenario: an invitation printing studio gets the same early questions: the date, the guest count, the paper stock, the venue style. Those questions point to the pages the same buyers read before they ever contact the shop.

Where context stops being the whole story

The Google Ads Help page on contextual targeting describes the process as matching ads to relevant sites on the Display Network by using factors such as your keywords or topics. It also lists language and location targeting and a visitor's recent browsing history among the factors. So Google's version is not a pure page-only method. If you want page-only placement, say so to the publisher and get their answer in writing.

How Google Ads describes the matching

According to the Google Ads Help page on contextual targeting (Opens in a new tab), the system analyses the content of each webpage to determine its central theme and then matches that theme to your ad. The page states that Google Ads uses contextual targeting when an ad group has keywords or topics and its campaign is set to show ads on the Display Network.

The same page says ads show on content that matches any of the topics, placements, or Display, Video, or Search keywords you target. Its own example: if you target a "bikes" topic and a "cycling" keyword, ads can show on content matching either one. In other words, the tools are additive. Adding a keyword does not switch the topic off.

Direct buys from publisher sites work differently from a Google Ads account, and the setting names do not carry over. What carries over is the logic: describe the subject, check the real pages, and set limits.

The IAB Tech Lab Content Taxonomy as a shared vocabulary

An advertiser may say "wedding paper goods," a publisher may say "bridal planning," and an agency may say "events and celebrations." Three people, three words for one subject. The IAB Tech Lab Content Taxonomy page (Opens in a new tab) presents the taxonomy as a shared language for describing content and names contextual targeting and brand safety as its uses.

For you, that means a brief can carry two descriptions of the same subject: your own phrase and a taxonomy category name. If the other side uses the taxonomy, misunderstandings drop. If they do not, the written category name still gives both sides something concrete to discuss. The documentation has no statement making the taxonomy mandatory for any publisher or platform.

What the version history says

The IAB Tech Lab page lists version 3.1 as released in December 2024, with an update to the genre category to include Thriller. It lists version 3.0 as released in June 2022, updated to better support news, video and connected TV content, podcasts, radio, games and app stores. According to the page, the 3.0 taxonomy expands coverage from roughly 400 categories to more than 1,500, and 2.x is being deprecated. The page says version 1.0 is deprecated and should not be used for content categorization in the absence of the flags it mentions. No separate category count for 3.1 appears on the page, so none is stated here.

Version details change; check the current IAB Tech Lab page before you rely on a number.

When an old publisher label meets a new category name

The page says 3.0 includes some deletions, so it is not backwards compatible with earlier releases. A publisher who tagged pages under 2.x may have codes that do not map one to one onto the 3.x names in your brief. The page mentions an open-source mapper, donated by Mixpeek, that converts 2.x category codes into 3.0 equivalents using several methods, from exact label match to LLM re-ranking, and returns codes, labels, confidence scores and methods. For a small campaign you do not need to run it. Asking the publisher which version they tag with is enough, and a low confidence score on a mapped label is a reason to ask for sample pages.

Building a topic ladder for the brief

Practical suggestion: build a ladder with three rungs. The bottom rung is the subject itself, the middle rung holds subjects readers study before they buy, and the top rung holds subjects that come right after.

The ladder keeps the brief honest. Each rung needs its own message, because a reader choosing a venue wants different information than a reader chasing late RSVPs. Copying the same ad across all rungs wastes the context you paid for.

Reading sample pages

A title that contains your keyword proves little. Open the page and read the first paragraph and the section headings. Ask three questions: is the page's main theme the one on your ladder, is the tone neutral or positive toward your subject, and does the page invite action, such as planning or buying? Mark each page as direct, neighbouring, distant or unclear.

Choosing between topics, keywords and placements

The Google Ads placement targeting page (Opens in a new tab) defines placements as locations where ads can appear on YouTube or the Google Display Network, and says you can hand-pick them. Placements can be set in the Content section while creating a campaign, or edited later from the Content tab in the left-side menu. If you use the interface in Turkish, label names may look different. According to the page, a placement must be part of YouTube or the Display Network for ads to show there, and if a targeted website has an equivalent app, ads can show there too.

Practical suggestion: treat the three tools as a progression.

The control and workload judgements in the table are practical suggestions.

When you buy directly from a publisher, the logic resembles a placement: you pick the site. The difference is that the deal sits between you and the publisher, not inside Google's network. A topic label on a publisher card is a starting point, not evidence. For a wider view of vetting, see the publisher quality checklist.

From label to real page: checking a publisher

Ask for page addresses, not slogans. A publisher who says "our whole site is about weddings" has not answered a page-level question. Ask for the section names, three example addresses per rung of your ladder, and the way the publisher decides the main theme of a mixed page. According to Google's page, the system analyses each webpage to determine its central theme, so Google thinks at page level too, and your review should match that level.

Page-level and site-level fit

A site's overall subject and a single page's subject can differ. A general lifestyle site may carry one excellent invitation-etiquette article among hundreds of unrelated posts. A wedding magazine may run a gossip section that does not suit an invitation printing studio. Write down which sections you want, and ask for section addresses in writing.

Questions to put to the publisher

• Which taxonomy and which version do you tag pages with?
• Can you share three example addresses for each topic on my list?
• Are topic labels assigned automatically, or does an editor approve them?
• How do you decide the main theme of a mixed page?
• Can you apply an exclusion list, and at which level: page, section or site?
• Does the site have a mobile app, and would my ad appear there?
• Will the final report show results by topic or section?

The last question matters for learning. If the report cannot split results by topic, you cannot tell which subject worked, and the next campaign starts blind. See how to measure banner ad performance for the reporting side.

Brand safety and exclusions

The IAB Tech Lab page names brand safety as one use of the taxonomy. Brand safety means your ad does not appear beside content you do not want to be associated with. It has two halves: choosing the subjects you want, and writing down the ones you refuse.

According to the Google Ads page, exclusions are one of the content targeting types managed on the same Content page as topics, placements and keywords. The documentation has no statement on how exclusions work, so look at Google's dedicated exclusion documentation for the mechanics.

Writing the exclusion list

For an invitation printing studio, sensitive adjacent subjects might include stories of cancelled weddings, bitter family disputes or heavy news coverage. That decision belongs to the business: some brands are comfortable beside a candid story, others are not. Practical suggestion for building the list:

1. Write three to five subjects your brand will not sit beside.
2. Ask the publisher at what level the list is applied.
3. Put the list into the written brief and date it.
4. Ask for a sample screenshot after launch.
5. Update the brief whenever the list changes.

When a topic and an exclusion collide

A page about invitation wording that also tells a story of a cancelled wedding creates a conflict between the topic you picked and the subject you excluded. Ask the publisher which rule wins. Practical suggestion: let the exclusion win, because a brand-safety slip costs more than a missed impression.

Example scenario: an invitation print shop before the autumn wedding season

Example scenario: Maple Lane Press is a small shop that prints wedding invitations and save-the-date cards. The owner wants banner ads on two publisher sites ahead of the autumn season, aiming to send readers to the page with the sample-pack request form.

Step one is the subject phrase: "custom wedding invitations and printing." Step two is the ladder: venue selection and budgeting before the purchase, guest list and RSVP handling after it. Step three is to ask both publishers for three sample pages per rung.

The first publisher is a wedding planning magazine. Its "how to choose a venue" article sits on the before rung and counts as a neighbouring match. Its "invitation wording" guide is a direct match. A gossip section about celebrity weddings looks distant, and since the owner prefers a calm brand tone, it goes on the exclusion list. The second publisher is a local lifestyle site with a mixed events section. Two of the three sample pages turn out to be roundups of many event types, so they are marked unclear and the owner asks for more samples before committing.

Step four is messaging. In the planning magazine, the "Plan the paper early" message goes beside the venue article, and a "Wording ideas for formal invitations" message goes beside the wording guide. Step five is tagging each placement with its own tracking value so the report separates the rungs.

The scenario shows that context is a decision process with several checkpoints, not a single label.

Second round of the same scenario

Example scenario, continued: after the first season the owner opens the ladder and writes an observation beside every rung: worked, did not work, not enough data. Sample-pack requests came from the wording guide section. The venue pages brought clicks without requests, so the owner does not drop them yet; the sample is thin, so the same subject runs again with a different message. The gossip exclusion held, confirmed by a screenshot from the publisher. For the next round the larger share of the budget moves to the section that worked, and a small share goes to a new neighbouring subject, the guest list.

Decision rule: topics first or a named placement

Decision rule: start with topic targeting if two conditions hold. First, your budget can be split across several publishers. Second, the publishers can confirm topic labels in writing. If both hold, start with topics and look at the first report to see which publisher and which page group produced requests. If the budget is small and you already trust one publisher, buy placements on that publisher's direct-match sections. If a publisher cannot supply sample pages, do not rely on its topic labels.

This is a practical suggestion. The Google Ads pages describe the targeting types, and the documentation has no statement on a decision rule between them.

Testing the context hypothesis

Contextual targeting is a hypothesis: readers on venue-planning pages are closer to ordering invitations than readers on general lifestyle pages. To test it, give each subject its own link and its own tracking value, so that the report shows which subject sent the visit. Practical suggestion: change only one thing at a time, either the subject or the message.

Two measuring traps are worth naming. The first is reading too much into a small number of impressions, where a difference in clicks can be chance. The second is mixing the publisher's report with your own analytics, since two systems can count the same event differently. For checking whether impressions were actually visible, see ad viewability and IAB standards. If you suspect bad traffic, the invalid traffic and click verification guide covers the checks.

Decision table: reading results by subject (practical suggestion)

Limits of contextual targeting

Four limits are worth keeping in mind. First, theme detection is an analysis step. The Google page says the system analyses the page to find its central theme, and nothing on the page says the analysis never errs. A mixed page can be matched to the wrong theme, which is why reading sample pages by hand beats trusting automatic matching.

Second, pages on one subject can differ in tone. An invitation article might be a how-to guide, a complaint about a printer, or a story of a mix-up at a wedding. The label does not show the difference. Reading the page does.

Third, the Google Ads page records a restriction: content targeting can no longer be added to new or existing video campaigns that drive conversions, and from early 2023 existing content targeting settings were removed automatically from those campaigns. The restriction concerns a specific Google Ads campaign type and says nothing about direct publisher buys. If you cannot find content targeting in your account, check the campaign type.

Fourth, context alone does not decide results. The creative, the offer and the landing page matter as much. A perfectly matched page still loses the order if the sample-pack form is slow or the delivery time on the page is out of date.

Agency and ecommerce notes

For an online store, topic choice starts from product categories, but the decision pattern is the same: direct pages, neighbouring pages and distant pages. For landing page and stock alignment, see ad space planning for ecommerce campaigns.

Agencies face a different problem: several clients' subjects pile up in one publisher list. Practical suggestion: keep one ladder per client, send the publisher subject-based briefs, and add exclusion lists only with the client's approval. Read the general buying steps in how to buy ad space on websites alongside this process.

Checklist for a contextual brief

• The ad subject is written as a short phrase.
• At least three neighbouring subjects are on the ladder, each with its own message.
• Subjects are named with a shared vocabulary such as the IAB Tech Lab taxonomy, with the version noted.
• Three sample page addresses per subject have been received, opened and read.
• Each sample page is marked direct, neighbouring, distant or unclear.
• The exclusion list is written and shared with the publisher.
• Each subject has its own tracking value and landing page.
• The publisher confirmed that the final report splits results by subject.
• Mobile app coverage and the rule for mixed pages are written down.

Misreadings to avoid

Treating the keyword in a title as proof

A title with "invitation" in it may belong to a complaint thread, a news item or a sale listing. Read the main theme.

Skipping the taxonomy version

If the publisher tags with 2.x and your brief uses 3.x names, the matches can slip. Add the version number to the brief.

Reading context as an audience list

Contextual targeting does not hand you a list of people. The reader on the page may or may not be your buyer, so set expectations accordingly.

Leaving exclusions until later

If a brand-safety problem is noticed after launch, those impressions have already run. Write the list before launch.

Frequently Asked Questions

Is contextual targeting the same as keyword targeting?

No. According to the Google Ads Help page, the two can be used in the same campaign. A topic is a content category, a keyword is a phrase, and your choice depends on how much control you need.

Do I have to use the IAB Content Taxonomy?

No. The IAB Tech Lab presents the taxonomy as a shared language, and the documentation has no statement on any obligation to use it. If your publisher uses another vocabulary, put your own subject phrase in the brief as well.

How many categories does the taxonomy have?

According to the IAB Tech Lab page, version 3.0 expands coverage from roughly 400 categories to more than 1,500. The documentation has no statement on a separate count for 3.1.

Do placements include mobile apps?

According to the Google Ads page, if a targeted website has an equivalent app, ads can show there too. In a direct publisher deal, ask the publisher about app coverage separately.

Does an exclusion list block everything on the list?

The documentation has no statement on how exclusions work. Agree the level of enforcement with the publisher in writing and ask for a sample screenshot.

Is contextual targeting better without cookies?

Neither the IAB nor the Google page makes that comparison. The Google Ads page also lists a visitor's recent browsing history among the factors it considers.

Should a small budget start with topics or placements?

Practical suggestion: if you trust one publisher, buy placements on its direct-match sections. If you know no publisher, run a small topic test first and move to placements based on the results.

Conclusion

Contextual targeting ties an ad to the page being read, not to the person. It works when you name the subject clearly, open the publisher's real pages, and put your brand-safety limits into a written brief. The IAB Tech Lab Content Taxonomy supplies a shared vocabulary for that work, and the Google Ads Help pages show how topics, keywords and placements fit together in one account.

Once your subject ladder is ready, browse the ad spaces and shortlist the publishers whose sections match it.