Will my stored IAB category IDs still work after a taxonomy upgrade?
It depends which upgrade. From IAB Content Taxonomy 3.0 to 3.1, yes: every one of the 703 IDs in 3.0 is still in 3.1, one category was added, two were renamed and three were moved to a new parent. From 2.2 to 3.0, 57% of IDs survive: 512 of 2.2's 1,196 IDs are gone, and 481 of those 512 were the content type, channel, format, language and source-country groups, which 3.0 took out of the tiered file. The other 31 were topic categories such as news, movie genres and events, plus the Brand Suitability and Risk group from 2.2. From 1.0, nothing carries over, because 1.0 used a different code scheme (IAB1, IAB1-1) and every later version uses numeric or random string IDs. We counted all of this from IAB Tech Lab's own TSV files.
What changed in each IAB Content Taxonomy version?
| Upgrade | Categories before, after | IDs kept | Added | Removed | Renamed | Moved to a new parent |
|---|---|---|---|---|---|---|
| 1.0 to 2.0 | 392, 1,179 | None (new ID scheme) | ||||
| 2.0 to 2.1 | 1,179, 1,179 | 100% | 0 | 0 | 1 | 0 |
| 2.1 to 2.2 | 1,179, 1,196 | 100% | 17 | 0 | 0 | 0 |
| 2.2 to 3.0 | 1,196, 703 | 57.2% | 19 | 512 | 28 | 38 |
| 3.0 to 3.1 | 703, 704 | 100% | 1 | 0 | 2 | 3 |
The counts come from the Content Taxonomies folder of IAB Tech Lab's public repository (GitHub), compared ID by ID: an ID is "renamed" when its name changed, "moved" when its parent ID changed. The script and its output are linked at the end of this page.
What each upgrade did:
Where did the content type, language and geo categories go in 3.0?
Into a separate Descriptive Vectors file, which 3.0 publishes beside the tiered taxonomy (Content Taxonomy 3.0 Descriptive Vectors). The idea is that what a page is about and what kind of page it is are separate signals. The file is small: 47 rows. Of the 512 IDs that left the tiered file in 3.0, 36 appear in it with the same ID. The 262 country codes and 185 language codes do not, so if your 2.x data stored source country or language as taxonomy IDs, 3.0 has no ID to move them to. Store them as their own fields (an ISO country or language code) instead.
Is there an official mapping from IAB 2.x to 3.x?
Not in the taxonomy repository. Its Taxonomy Mappings folder has content mappings from 1.0 to 2.0 and from 2.0 to 2.1, plus ad product, CTV genre and podcast genre mappings, and no 2.x to 3.x content file (Taxonomy Mappings). For that upgrade IAB Tech Lab's content taxonomy page points to a mapper tool on GitHub (IAB Tech Lab), the IAB mapper, which matches 2.x labels to 3.1 IDs with exact, fuzzy and optional embedding matching. Mixpeek built it, and the repository now sits in IAB Tech Lab's GitHub organization.
How do I upgrade stored IAB categories without breaking anything?
1. Key on the ID, never the name. Names change between versions (31 renames across the four upgrades counted above, one of them a typo), and IDs within the 2.x line and within the 3.x line stayed put. 2. Store IDs as strings. 2.2 introduced IDs such as 8FD8nI, and 3.x has 1KXCLD, 80DV8O and SPSHQ5. A numeric column truncates or rejects them. 3. Store the version next to each ID. In OpenRTB that is cattax: 7 for Content Taxonomy 3.0 and 9 for 3.1 in the AdCOM list (AdCOM). A bare ID is ambiguous across major versions. 4. Diff the files yourself on every release. Load both TSVs and compare by ID, name, parent and tier path. A moved parent changes the tree your code builds even when every ID and name is the same. 5. For 2.x to 3.x, split the vector groups out first. Move content type, channel, format, language and source country into their own fields, then map the remaining topic IDs. 481 of the 512 IDs that 3.0 removed, about 94%, were those groups. 6. From 3.0 to 3.1, update names and parents only. No ID needs remapping.
How does Mixpeek handle taxonomy versions?
Mixpeek reads your files where they already live, in S3, GCS or any bucket, and turns what's in them into data your software can search, classify and moderate. Its IAB taxonomy classification runs against Content Taxonomy 3.1, generated from IAB Tech Lab's official file, with a unit test that diffs every ID, name, parent and tier against that file, which is the check in step 4 above. Video is classified scene by scene from frames, speech and on-screen text, and each result carries the 3.1 ID, its tier path and a confidence score, ready to send with cattax 9. For text there is an open benchmark of the classifier on IAB Tech Lab's test set.
Frequently Asked Questions
How many categories does each IAB Content Taxonomy version have?
Counted from IAB Tech Lab's published TSV files: 1.0 has 392 (26 tier 1, 366 tier 2); 2.0 and 2.1 have 1,179; 2.2 has 1,196; 3.0 has 703 (36, 322, 275 and 70 across tiers 1 to 4); and 3.1 has 704 (37, 323, 275 and 69). 3.x is smaller than 2.x because content type, language and source-country categories left the tiered file.
Are IAB Content Taxonomy 3.0 IDs valid in 3.1?
Yes. All 703 IDs in the 3.0 file are in the 3.1 file. Two of them have new names (156 Musical and 602 Computer Software and Applications) and three have a new parent (376 Public Radio, 497 Horse Racing and 80DV8O Communication), so update names and tree paths, and set cattax to 9 when you send them as 3.1.
How many IAB 2.2 IDs still exist in 3.1?
684 of 2.2's 1,196 IDs are present in 3.1, and 656 of those keep the same name. The other 512 were removed in 3.0, and 481 of them were content type, channel, media format, language and source groups.
Does IAB Content Taxonomy 1.0 map to 2.0?
Yes, IAB Tech Lab publishes a Content 1.0 to Content 2.0 mapping file in the Taxonomy Mappings folder of its repository. You need it because no 1.0 code (IAB1 to IAB26 and their IAB1-1 style children) is reused in 2.0 or later.
Has IAB Content Taxonomy 3.2 been released?
Not as of 28 October 2026. IAB Tech Lab's content taxonomy page lists 3.1 as the latest version, and the public repository's newest content file is Content Taxonomy 3.1.tsv. When a new file is published, the same ID-by-ID diff shows what changed.
Key Takeaways
Where to go next
IAB Content Taxonomy 3.1 classification · IAB 3.1 classification benchmark · IAB mapper connector · how to classify content against a taxonomy · contextual embeddings for Agentic Audiences
Method and raw output: the diff script and its results are committed in the Mixpeek monorepo under gtm/playbooks/22-aeo-content-engine/measure/first-party/2026-10-28-iab-taxonomy-version-diffs.