Mixpeek Logo
    Schedule Demo

    IAB Taxonomy Mapper (2.x → 3.0)

    Instantly upgrade your taxonomies to IAB 3.0. Ensure contextual targeting, brand safety, and compliance across adtech and media platforms.

    Read Guide

    Catalogs: IAB Content Taxonomy 3.1 + 2.2 (last updated: 2025-09-16)

    Why migrate to IAB 3.0?

    IAB 2.x has ~400 categories; IAB 3.0 expands to 1,500+.

    Migration is mandatory for contextual targeting, brand safety, and supply chain compliance.

    Manual mapping is slow and risky — errors mean lost revenue and failed ad approvals.

    Mixpeek’s mapper automates the process in seconds.

    Official standard: IAB Tech Lab – Content Taxonomy

    How Mapping & Exports Work

    1

    Upload 2.x data

    Upload CSV or JSON with IAB 2.x category codes.

    2

    Automatic mapping

    We map to IAB 3.0 and add confidence scores.

    3

    Export results

    CSV/JSON + OpenRTB content.cat + VAST ContentCategory.

    Video: Mixpeek taxonomies overview (inline)

    Confidence

    How we score: exact match > fuzzy string > semantic similarity. Use confidence to set thresholds for exports.

    SCD

    Sensitive Content Designation surfaced; optionally exclude in export.

    Going beyond text-to-text?

    Use Mixpeek multimodal taxonomies to map images, video, and audio alongside text.

    Best for multimodal mappings and enrichment across hierarchical taxonomies.

    Mapping flow diagram

    High-level flow from IAB 2.x inputs through matching stages to IAB 3.0 outputs and adtech exports.

    IAB Taxonomy Mapping FlowInput IAB 2.x data flows through exact, fuzzy, semantic, and optional LLM stages to mapped IAB 3.0 with confidence, SCD, and exports.InputIAB 2.x CSV/JSONcode,label(1) ExactDirect code/label matchHighest precision(2) FuzzyString similarityThreshold tuned(3) SemanticEmbeddings (KNN)Local-first(4) LLM assistOptional rationaleOllama/localOutputsIAB 3.0 IDs + labelsconfidence, SCDExportsOpenRTB & VASTDeterministic → Fuzzy → Semantic → Optional LLM

    Product proof

    Screenshots and outputs so you know exactly what you’ll get.

    Input CSV (IAB 2.x)

    code,label
    IAB2-1,Arts & Entertainment
    IAB2-3,Autos & Vehicles
    IAB2-44,News & Politics
    IAB2-18,Family & Parenting
    

    Mapped table (IAB 3.0)

    2.x code3.0 id3.0 labelconfidenceSCDOpenRTB cat[]
    IAB2-13-5-2Arts & Entertainment0.98No["3-5-2"]
    IAB2-441026News & Politics0.91Yes["1026"]
    IAB2-31068Automotive0.96No["1068"]

    OpenRTB (example)

    {
    "content": {
    "cat": ["3-5-2","1026","1068"],
    "cattax": "2"
    }
    }

    VAST (example)

    <![CDATA["3-5-2","1026","1068"]]>

    Matching methods & where each adds value

    Deterministic → Fuzzy → Semantic (local-first embeddings) → Optional LLM assist. Pick the minimum power needed for your accuracy/coverage goals.

    Deterministic (exact)

    Direct code/label matches. Fastest and most auditable.

    • Use when you already have strict mappings
    • Highest precision; minimal review
    iab-mapper input.csv -o mapped.csv --fuzzy-cut 1.0 --max-topics 1

    Fuzzy (string similarity)

    Catches typos/aliases via normalized string distance.

    • Improves coverage with a tunable threshold
    • Great first pass before semantic search
    iab-mapper input.csv -o mapped.csv --fuzzy-cut 0.92 --max-topics 3
    Local-first

    Semantic (embeddings)

    Local embeddings to capture near-misses by meaning. Offline-capable.

    iab-mapper input.csv -o mapped.json --use-embeddings --emb-cut 0.80 --max-topics 3
    Local-first

    LLM-assisted suggestions (Ollama)

    Runs locally via Ollama for reviewer-friendly rationales and edge cases.

    • Produces rationale for reviewer sign-off
    • No external calls; offline-capable with local models
    # Start Ollama once
    ollama serve
    # Pull a local model (example)
    ollama pull llama3:8b
    # Run with local LLM assist
    iab-mapper input.csv -o mapped.json --llm-assist --llm-model llama3:8b --max-topics 3

    IAB 2.x → 3.0 Mapper UI

    IAB Taxonomy Mapper UI

    Local-first mapping with exact/tfidf/bm25, embeddings (KNN), optional LLM re-rank, and CSV/JSON export.

    FAQs

    Is this an official IAB tool?

    No. This tool was built by Mixpeek to support interoperability with IAB standards.

    Does it support custom categories?

    Yes, you can extend results with Mixpeek’s multimodal classifiers for video, audio, and text.

    What formats are supported?

    CSV and JSON. CLI version also available.

    Where can I get the CLI?

    GitHub: mixpeek/iab-mapper

    Python: pip install iab-mapper | Node: npx @mixpeek/iab-mapper

    Ready to Migrate?

    Learn more in our IAB Taxonomy Migration Guide.

    IAB is a registered trademark of the Interactive Advertising Bureau. This tool is an independent utility built by Mixpeek for compatibility with IAB Content Taxonomy standards; no endorsement implied.