AI Product Found Rate

Found Rate is the percentage of relevant shopping queries on which a retailer's product appears in the AI engine's response in any form - as a named brand mention, as a product card, as a citation, or as a recommendation. It is the baseline commerce visibility metric for ACO programs, analogous to organic impression count in traditional SEO.

Found Rate sits alongside two other commerce visibility metrics:

• Found Rate - does the product appear at all?
• Product Card Rate - does the product appear with image, price, and link (not just text)?
• Visibility Score - how prominently does it appear (position in the answer, primary recommendation vs. alternative)?

All three are measured per engine, per query panel, over time. A product with high Found Rate and low Product Card Rate is being mentioned as a brand but not as a shoppable product - usually a schema or feed gap. A product with high Found Rate and high Product Card Rate but low Visibility Score is being included as an alternative rather than a top recommendation - usually a review content or category authority gap.

A practical Found Rate measurement program runs on four decisions:

1. Query panel selection. 30-100 queries per category. Mix branded ("is [your brand] good for X?"), semi-branded ("[your brand] vs [competitor]"), and unbranded ("best X for Y"). Keep the panel fixed so results are trendable. Rotate a small subset quarterly to reflect new buying scenarios.

2. Engine coverage. ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Gemini, Copilot, and Amazon Rufus for categories where Amazon is relevant. Found Rate varies sharply by engine (Superlines measured a 14.8x sentiment gap between Perplexity and ChatGPT on a single brand), so per-engine tracking is required.

3. SKU-level granularity. Track Found Rate per SKU for hero products and per product line for the long tail. Brand-level Found Rate hides the fact that a retailer's top 20% of SKUs by revenue often appear well and the bottom 80% are invisible.

4. Weekly cadence. AI engines update retrieval indexes frequently. Monthly is too slow to react to competitor moves or your own content launches. Weekly is the floor.

Platforms like Paz.ai automate the panel execution and parsing across engines - running the same queries, extracting brand and product mentions, scoring Found Rate, Product Card Rate, and Visibility per engine per SKU. Manual testing works for small panels but does not scale past 20-30 queries across 5 engines weekly.

The levers map directly to ACO and AEO fundamentals:

1. Attribute depth. Enrich the catalog to 25-40 structured attributes per product (material, fit, use case, compatibility, certifications). This is the single largest Found Rate lever for retailers shipping thin catalogs.
2. Complete product schema. BrightEdge measured pages with complete product schema are 2.5x more likely to be cited in AI Overviews. Schema is table-stakes for Product Card Rate specifically.
3. Feed coverage across surfaces. Google Merchant Center (for UCP/Gemini/AI Overviews), ChatGPT Shopping feed (for ACP), category-specific feeds for Rufus and Perplexity. Running only one feed caps Found Rate at the share of queries that engine handles.
4. Category authority content. For unbranded queries (the majority of shopping intent), Found Rate depends on category authority more than on individual product optimization. Review guides, comparison content, and inclusion in third-party roundups all lift unbranded Found Rate.
5. Third-party review volume. AI engines heavily weight review sentiment and volume when composing recommendations. Paz's own analysis in its branded vs nonbranded visibility research found review volume was one of the top three predictors of unbranded Found Rate lift.
6. Unblocked AI crawlers. If GPTBot, Google-Extended, ClaudeBot, or PerplexityBot are blocked in robots.txt, Found Rate on the blocked engine is effectively zero regardless of everything else.