How does the IAB Content Taxonomy work, and how do planners map a brief to its IDs?
Part of Contextual targeting
The short answer
The IAB Content Taxonomy is IAB Tech Lab's standard list of content categories, each with a unique ID and arranged in up to four tiers, used to describe what a page, app or video is about. Planners map each brief topic to the deepest category that fits, record the version, and send those IDs to partners.
Key takeaways
- Content Taxonomy 3.1 is a tree of categories with up to four tiers, and each category has a unique ID that partners exchange instead of names.
- Version 3.x is not backward compatible with 1.0 or 2.x, so every list of IDs needs its version recorded beside it.
- Vectors describe purpose, source, format and environment separately from the subject, which is how news, reviews and how-to content are labeled.
- In OpenRTB the cattax field says which taxonomy the IDs come from, and when it is missing Content Taxonomy 1.0 is assumed.
- A good brief mapping goes to the deepest tier that fits, adds the right vectors and is tested against real pages before it is sent.
What is the IAB Content Taxonomy, in plain terms?
Definition
- IAB Content Taxonomy
- A published hierarchy of content categories, maintained by IAB Tech Lab, that lets publishers, exchanges, buying platforms and verification partners describe the subject of a page, app or video with the same IDs.
Tech Lab's implementation guidance (opens in a new tab) calls it a common language for describing content, and names two main uses: contextual targeting, which it also calls addressability, and brand safety and suitability. The taxonomy describes what content is about. It does not describe the person reading it, and it says nothing about whether a page is well made or whether an ad will perform.
The same Tech Lab working group also maintains two sister lists that are easy to confuse with it. The Ad Product Taxonomy describes the product being advertised in a creative, and the Audience Taxonomy names audience segments. A planner describing where an ad should appear works with the Content Taxonomy; a publisher deciding which ads to block works with the Ad Product Taxonomy.
Which versions exist, and which one should a plan use?
Tech Lab's standards page (opens in a new tab) lists 3.1 as the newest version and says Content Taxonomy 1.0 should not be used for content categorization in the absence of SCD flags. Its versioning policy states that 3.0 may not be used alongside versions 1.0 through 2.2, and the implementation guidance explains that 3.0 removed parent categories. Each version also has its own value in the AdCOM list that OpenRTB uses to say which taxonomy a set of IDs belongs to.
| Version | AdCOM cattax value | Status and notes |
|---|---|---|
| 1.0 | 1 | Deprecated; AdCOM recommends it not be used because it has no SCD flags. IDs look like IAB1-1. |
| 2.0 | 2 | Deprecated; AdCOM gives the same reason as for 1.0. |
| 2.1 | 5 | Introduced the SCD flag and extra vectors, such as the geographic origin of content and the channel where it is consumed. |
| 2.2 | 6 | Added brand safety and suitability support, with risk levels for the core safety categories. |
| 3.0 | 7 | Breaking change: parent categories removed, better coverage for video, news, podcasts, games and apps. |
| 3.1 | 9 | Newest version; Tech Lab notes an update to the genre category to include Thriller. |
How are tiers, parents and unique IDs organized?
Every row in the 3.1 file has a Unique ID, the ID of its Parent, a Name and its full path from Tier 1 to Tier 4. In our October 2026 count the file held 37 Tier 1 categories, 323 at Tier 2, 275 at Tier 3 and 69 at Tier 4. Many IDs are plain numbers kept for backward compatibility, while newer rows use short codes, such as SPSHQ5 for Genres or v9i3On for Sensitive Topics. The guidance says IDs, not names, should be used when tagging content.
| Tier | Name | Unique ID | Parent ID |
|---|---|---|---|
| Tier 1 | Personal Finance | 391 | none |
| Tier 2 | Personal Debt | 405 | 391 |
| Tier 3 | Home Financing | 407 | 405 |
| Tier 3, sibling | Student Loans | 409 | 405 |
Definition
- Tier
- The depth of a category in the tree. Tier 1 is the broad subject, such as Pets; Tier 4 is the narrowest, such as Menopause under Medical Health, Diseases and Conditions, Endocrine and Metabolic Diseases.
What are vectors and SCD flags, and why do planners need them?
Categories describe the subject. Vectors describe everything else about the content, and they sit in their own list. The guidance calls them orthogonal content attributes, which means a page carries a subject category and any number of vectors at the same time.
| Vector group | Example values | Why a planner cares |
|---|---|---|
| Content Purpose | Informational, with News, Opinion & Op-Ed, Review, Instructional; Entertainment; Commerce | Separates a how-to guide from a news report on the same subject |
| Content Source | Professionally Produced, User Generated, Aggregated/Curated | Flags forums and aggregators that need extra review |
| Content Form Factor | Textual, Video (Show, Clip, Movie, Event), Audio (Podcast, Radio), Game | Matches the brief's format to the inventory |
| Content Environment | General, Social, Forum/Community, Marketplace/eCommerce | Keeps store and listing pages apart from editorial pages |
| Brand Suitability and Risk | Floor, High Risk, Medium Risk, Low Risk | Expresses a risk level beside a sensitive topic |
News is the clearest case. The guidance says version 3.0 removed the old News and Politics subject category so that news is not confined to one branch, and recommends tagging news with the News or Opinion purpose vector plus any subject category, such as Business and Finance for business news.
Definition
- SCD flag (Special Category Data)
- A marker on categories that could be used to infer sensitive traits such as health, religion or politics if they were stored against a person's ID. In our October 2026 count, 63 rows in the 3.1 file carried it, including most Medical Health conditions and the Religion and Spirituality branch.
How do planners map a media brief to category IDs?
Seven steps from brief topics to a checked ID list
- List each topic in the brief as a subject. Separate subject from treatment: small business bookkeeping is a subject, a how-to guide is a treatment.
- Find the Tier 1 branch, then walk down. Open the 3.1 file, filter by the Tier 1 name and read every child before choosing.
- Stop at the deepest category that still fits. A deeper ID is more precise, but if the brief covers the whole branch, the parent is the honest choice.
- Add vectors for treatment and format. Use Instructional or Review for advice content, and Textual or Video for the format the creative needs.
- Map the avoid topics the same way. Use the Sensitive Topics children and any subject categories the brand wants kept away, with the risk level it tolerates.
- Write down the version and the IDs, not just names. A note such as Content Taxonomy 3.1, cattax 9 prevents a partner from reading the IDs against an older list.
- Test the mapping on real pages. Pull ten pages the brand would love and ten it would refuse, and check that the IDs separate them.
Fictional example
A bookkeeping app brief for Florida
An unnamed bookkeeping app for sole proprietors and small firms wants display and video ads in Florida, in English and Spanish, beside practical guides on running a small business and handling self-employment taxes. It wants to stay away from coverage of financial crises and bankruptcies.
The planner maps the subjects to Small and Medium-sized Business (60) and Business Accounting & Finance (54) under Business and Finance, Business, and to Personal Taxes (415) under Personal Finance. Startups (61) is added as a test topic. The treatment is the Instructional purpose vector, with Textual and Video form factors.
Avoid topics map to Financial Crisis (83) under Economy and Bankruptcy (65) under Business Banking & Finance. Because the category names are published in English, Spanish pages are tagged with the same IDs, and a Spanish reader checks a sample of matched pages before the list is sent.
How do taxonomy IDs travel in an OpenRTB bid request?
In OpenRTB 2.6 (opens in a new tab), the Site object carries three category arrays: cat for the site, sectioncat for the current section and pagecat for the current page. A cattax field says which taxonomy those IDs come from, with values defined in AdCOM's list of category taxonomies (opens in a new tab). App and Content objects follow the same pattern.
How does PlacementGraph use Content Taxonomy 3.1?
What should a planner check before sending category IDs to a partner?
Taxonomy mapping check before a list leaves the agency
- Every ID on the list comes from one version of the taxonomy, and the version is written on the list.
- The partner confirmed which cattax value it expects for that version.
- Each brief topic maps to the deepest category that fits, and any parent level choice has a reason.
- Purpose and form factor vectors are listed beside the subject categories.
- Avoid topics are mapped to IDs, with the tolerated risk level where the partner supports it.
- Old 1.0 style IDs such as IAB1-1 were translated using Tech Lab's published mappings and then reviewed by hand.
- Sample pages in each language of the brief were checked against the mapping.
Frequently asked questions
What is the difference between the Content Taxonomy and the Ad Product Taxonomy?
The Content Taxonomy describes what a page, app or video is about. The Ad Product Taxonomy describes the product or service in an ad. Tech Lab says Ad Product is mainly used by publishers and SSPs to block categories of ads, so a buyer plans placements with Content IDs while a publisher filters incoming ads with Ad Product IDs.
Can a plan still use Content Taxonomy 1.0 IDs?
Tech Lab says 1.0 should not be used for content categorization and has been deprecated since 2020, and AdCOM recommends against it because it lacks SCD flags. Tech Lab publishes mappings from 1.0 to later versions, but its guidance warns that some mappings are imperfect, so review each mapped ID by hand.
Are the category names translated into French or Spanish?
Tech Lab's files publish the names in English, and the IDs are the same whatever the language of the content. The language of a page is described separately with the Content Language Extension in the vectors list. A bilingual plan therefore uses one ID list, with a reviewer who reads each language checking sample pages.
Does a category label tell a buyer that a page is brand safe?
No. A subject category says what the page is about. Safety and suitability are expressed separately, through the Sensitive Topics categories and the Brand Suitability and Risk vector, and they depend on how a partner classifies the page. Treat a label as a description to verify, not as a verdict on the content.
Sources
- Content Taxonomy (opens in a new tab)IAB Tech Lab
- Content Taxonomy 3.1 (TSV file) (opens in a new tab)IAB Tech Lab, Taxonomies repository on GitHub
- Content Taxonomy 3.0 Descriptive Vectors (TSV file) (opens in a new tab)IAB Tech Lab, Taxonomies repository on GitHub
- Taxonomies implementation guidance (opens in a new tab)IAB Tech Lab, Taxonomies repository on GitHub
- Taxonomies repository and versioning policy (opens in a new tab)IAB Tech Lab on GitHub
- AdCOM v1.0, List: Category Taxonomies (opens in a new tab)IAB Tech Lab on GitHub
- OpenRTB 2.6 (opens in a new tab)IAB Tech Lab on GitHub
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.