The AI Commerce Benchmark
We measure the AI discovery layer. Ourselves. A quarterly measurement of how AI engines represent ecommerce catalogs: which products they cite, which they miss, and where visibility is won or lost.
Overview
AI search is becoming a product discovery layer: shoppers ask an AI system what to buy, and the system answers with specific products. For retailers, the question is no longer whether their pages rank, but whether their products appear in those answers, with correct attributes, under the right queries.
The AI Commerce Benchmark measures that question at scale. Each quarter we sample a catalog of 12,500+ products, test 42,000 queries across four AI engines, and record which products are cited, how they are described, and which queries return nothing usable at all.
The Q3 edition is in preparation. The methodology below is public now; results are published only when the dataset behind them is complete.
Methodology
The measurement pipeline runs the same way every quarter:
- Query sampling: queries are drawn from the product-discovery space (comparison, use-case, and category questions), not branded or navigational queries.
- Product matching: every citation is matched back to a catalog entity (product, category, brand, or variant) using structured attributes.
- Citation extraction: answers are parsed for product mentions, attribute accuracy, and source links.
- Scoring: each product-query pair receives a visibility score (cited, described accurately, sourced), which feeds the GEO score per query.
- Refresh: the full dataset is re-run quarterly; changes between editions are published in Changes since Q2.
Results — 2026 Q3
Results are published when the edition is complete. AEOcode Research does not publish findings without the dataset behind them.
Changes since Q2
Changes between editions are published with each quarterly release.
Download
When an edition ships, the complete dataset and per-category breakdowns will be available to AEOcode customers through the platform.