A transparent, repeatable methodology for measuring how AI understands, evaluates, and trusts commerce.
The same signals observed above do not stay loose. Each one is classified into the measurement layer it actually belongs to.
This is not a checklist. It is classification.
Hover a layer to see what it measures, how it is evaluated, and why it carries the weight it does.
Hover or click a layer to see how it is measured.
AI vision systems now read product images directly, and a mismatched alt tag makes a photo invisible to the AI brain even though the AI eye can see it fine.
GPT-4o Vision API, HTML parser, structured data comparison.
Structured data is the primary language AI agents use to understand what a store sells, at what price, and how trustworthy it is.
Schema.org validator, JSON-LD parser, llms.txt fetcher.
A technically broken store is an invisible store: blocked crawlers and slow pages break AI navigation before content is ever read.
robots.txt parser, Google PageSpeed Insights API, heading tree analysis.
AI agents run a machine-readable trust check before routing a buyer or completing a purchase, not a design check.
Policy page crawler, agentic checkout flow tester, footer parser.
AI agents are starting to complete purchases, so a wrong price or stock status breaks the transaction and the trust that comes with it.
Live scraper and feed XML comparison, availability checker.
AI agents match stores to specific buyer intent, not keywords, and generic copy cannot match a specific request.
LLM semantic embedding comparison.
There is a real difference between a store AI mentions and a store AI trusts enough to surface with confidence.
Schema validator, sentiment analysis API.
A brand with no mentions or citations outside its own channels is treated as unverified by AI systems trained on the wider internet.
LLM citation indexing, Reddit API, Perplexity check.
Every layer is scored on its own scale, then weighted, normalized, and checked for confidence before it becomes the final AI Commerce Score.
Move across the scale to see how confidently AI can read a store at each point, from invisible to fully legible.
Move across the scale above to see the full example for that range.
A weight or a threshold is never assumed. Every part of this methodology is checked against real, ongoing evidence.
Every layer expanded: its submetrics, how each one is evaluated, the research behind its weight, and an example of what it catches. Hover any card to open it.
This methodology measures what can be observed from outside an AI system, not what happens inside one. No outside party has access to the internal weights or proprietary logic of ChatGPT, Gemini, Claude, or Perplexity, and any claim otherwise should be treated with skepticism. A high score means AI can read, understand, and trust a store. It is not a prediction of whether AI will recommend that store in a real conversation; measuring that directly is what Recommendation Intelligence is for.
Most methodologies stop at measurement. This one is built to run in a loop.
Find out exactly what AI agents see when they evaluate your store against every measurement layer described above.
Illustrative example · single-site signal for atomfoundry.dev.
View full signals →