AI Commerce Score™ Methodology

Every AI Commerce Score™ begins with observation.

A transparent, repeatable methodology for measuring how AI understands, evaluates, and trusts commerce.

example-store.com
Observing
SSL
HTTPS
Reviews
Schema
Shipping
FAQ
Pricing
Trust Badges
Open Graph
Structured Data
147 signals observed
From signals to measurement

Every signal belongs somewhere.

The same signals observed above do not stay loose. Each one is classified into the measurement layer it actually belongs to.

SSL
Technical
HTTPS
Technical
Reviews
Trust
Trust Badges
Trust
Schema
Structured Signals
Open Graph
Structured Signals
Structured Data
Structured Signals
Shipping
Commerce
Pricing
Commerce
FAQ
Intent

This is not a checklist. It is classification.

The eight measurement layers

One score, built from eight layers.

Hover a layer to see what it measures, how it is evaluated, and why it carries the weight it does.

AI Commerce Score
100
points total

Hover or click a layer to see how it is measured.

Semantic Visuals & Image Clarity15%

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.

Observed signals
Alt TextImage OverlaysVisual Consistency
Evaluation

GPT-4o Vision API, HTML parser, structured data comparison.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
AI Structured Signals15%

Structured data is the primary language AI agents use to understand what a store sells, at what price, and how trustworthy it is.

Observed signals
Schema.orgMerchant Identityllms.txtOpen GraphStructured Data
Evaluation

Schema.org validator, JSON-LD parser, llms.txt fetcher.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Core Technical & Interpretability15%

A technically broken store is an invisible store: blocked crawlers and slow pages break AI navigation before content is ever read.

Observed signals
SSLHTTPSrobots.txtCore Web VitalsDOM Structure
Evaluation

robots.txt parser, Google PageSpeed Insights API, heading tree analysis.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
AI Trust & Transaction Confidence15%

AI agents run a machine-readable trust check before routing a buyer or completing a purchase, not a design check.

Observed signals
ReviewsPoliciesContactReturnsCompanyShipping
Evaluation

Policy page crawler, agentic checkout flow tester, footer parser.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Commerce & Feed Accuracy15%

AI agents are starting to complete purchases, so a wrong price or stock status breaks the transaction and the trust that comes with it.

Observed signals
PricingInventoryShipping
Evaluation

Live scraper and feed XML comparison, availability checker.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
User Intent Match10%

AI agents match stores to specific buyer intent, not keywords, and generic copy cannot match a specific request.

Observed signals
FAQLong-Tail CopyContext Phrases
Evaluation

LLM semantic embedding comparison.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Recommendation Confidence10%

There is a real difference between a store AI mentions and a store AI trusts enough to surface with confidence.

Observed signals
Reviews SchemaSentiment
Evaluation

Schema validator, sentiment analysis API.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
External Authority Signals5%

A brand with no mentions or citations outside its own channels is treated as unverified by AI systems trained on the wider internet.

Observed signals
Brand MentionsCitations
Evaluation

LLM citation indexing, Reddit API, Perplexity check.

Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Watch the score take shape

Eight layers become one number.

Every layer is scored on its own scale, then weighted, normalized, and checked for confidence before it becomes the final AI Commerce Score.

0
Semantic
0
Structured
0
Technical
0
Trust
0
Commerce
0
Intent
0
Recommendation
0
Authority
Weighting
Normalization
Confidence
0/100
AI Commerce Score™
What the score means

Where does your number land?

Move across the scale to see how confidently AI can read a store at each point, from invisible to fully legible.

Low Confidence
0255075100
0 to 49
AI Invisible Risk
Critical gaps prevent AI from reading the store at all. It is being skipped today, not because it is a bad store, but because AI cannot verify it is a real one.
50 to 69
Low Confidence
AI can see the store but misses critical signals. What it does read is weak, inconsistent, or easily lost to a better-signaled competitor.
70 to 84
Moderate Confidence
AI can read and verify the store with reasonable confidence. Stronger signals in the weaker layers would raise how clearly and consistently AI understands it.
85 to 100
High Confidence
AI agents understand, trust, and can read the store with full confidence. Only 77 stores in the database have reached this threshold so far.

Move across the scale above to see the full example for that range.

How we validate

Checked, not just calculated.

A weight or a threshold is never assumed. Every part of this methodology is checked against real, ongoing evidence.

Controlled Experiments
Founder Lab
a live, permanent control store
Repeated Measurements
Same store.
Same methodology.
Large-scale Benchmarking
66,090
stores scanned
Continuous Updates
The methodology evolves as AI does
Transparency

The methodology, by the numbers.

3.1
Version
Jun 2026
Updated
8
Measurement layers
100
Scoring scale
66,090
Current database
Open documentation

The eight layers, in full.

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.

01Semantic Visuals & Image Clarity15%

Submetrics

Alt Text Matching8%
We send your product images to GPT-4o Vision and compare its semantic description against your actual alt text. If the image shows a black leather Chelsea boot and your alt attribute says product-image-4, that gap costs you points.
Tool: GPT-4o Vision API + HTML parser
Image Context & Overlays4%
We check whether critical information is baked into images as graphics rather than HTML text, invisible to most AI systems.
Tool: Vision API overlay detection
Product Visual Consistency3%
We verify that what appears in the product photo matches what the HTML description says.
Tool: Vision API + structured data comparison
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a product photo of a black leather boot with an alt tag reading product-image-4.
02AI Structured Signals15%

Submetrics

Schema.org Validation5%
We validate the presence and correctness of Product, Offer, and AggregateRating JSON-LD markup.
Tool: Schema.org validator + JSON-LD parser
Merchant Identity5%
We check for Organization or Store schema that connects your website to your Persistent Merchant Identity.
Tool: JSON-LD parser + knowledge graph check
AI Root Indexing5%
We check for a valid llms.txt file in your domain root, the file that tells AI agents what your store sells and how to navigate it.
Tool: llms.txt fetcher + content quality scorer
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a store with valid Product and Offer schema but no llms.txt file at its domain root.
03Core Technical & Interpretability15%

Submetrics

Crawler Accessibility5%
We check whether robots.txt and any WAF configuration block known AI crawlers including GPTBot, ClaudeBot, and AppleBot.
Tool: robots.txt parser + user-agent tests
Core Web Vitals5%
We pull Largest Contentful Paint and Cumulative Layout Shift via the Google PageSpeed API.
Tool: Google PageSpeed Insights API
DOM Structure Quality5%
We analyze the heading hierarchy and the ratio of semantic HTML to div soup in the rendered DOM.
Tool: HTML parser + heading tree analysis
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a Cloudflare rule silently blocking GPTBot from crawling the catalog.
04AI Trust & Transaction Confidence15%

Submetrics

Machine-Readable Policies5%
We check whether the return policy and terms of service are in crawlable HTML text, not locked inside a JavaScript modal or a PDF.
Tool: Policy page crawler + text extraction
Checkout Gateway Accessibility5%
We test whether the cart and checkout flow is accessible to autonomous agents.
Tool: Agentic checkout flow tester
Legal Identity Transparency5%
We check whether the company behind the store is clearly identified with a readable business name, address, and contact method.
Tool: Footer parser + identity verification
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a return policy locked inside a JavaScript modal that no crawler can read.
05Commerce & Feed Accuracy15%

Submetrics

Real-Time Price Sync8%
We compare the price visible in the HTML against the price in the Google Shopping XML feed or Shopify API.
Tool: Live scraper + feed XML comparison
Inventory Integrity7%
We check whether availability signals are accurate across the feed, the schema markup, and the live page.
Tool: Feed vs. live page availability checker
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a feed showing 89 dollars while checkout charges 109 dollars with a delivery fee.
06User Intent Match10%

Submetrics

Long-Tail Semantic Coverage10%
We use semantic embedding comparison to measure how well product descriptions cover the actual language buyers use in LLM search prompts.
Tool: LLM semantic embedding comparison
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a buyer asking for a gift for a dad who loves hiking under 60 dollars.
07Recommendation Confidence10%

Submetrics

Reviews Schema & Sentiment10%
We check for AggregateRating schema with a real ratingValue and reviewCount, and run sentiment analysis on accessible review content.
Tool: Schema validator + sentiment analysis API
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a store with real reviews but no AggregateRating schema exposing them to AI.
08External Authority Signals5%

Submetrics

Knowledge Graph Footprint5%
We check for brand mentions and citations in external LLM indexes, Reddit discussions, Perplexity citations, and AI-readable databases.
Tool: LLM citation indexing + Reddit API + Perplexity check
Research basis: weight and thresholds for this layer are benchmarked across the store dataset and refined through Atom Foundry's ongoing mechanism studies.
Example: a brand with strong reviews but zero mentions in Reddit threads or AI citations.
Limitations

What this methodology does and does not measure.

What this methodology measures
Observable AI behavior
Store readiness
Recommendation Confidence signals
What it does not measure
Internal model weights
Proprietary ranking systems
Hidden prompts
Whether AI actually recommends you in practice

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.

Continuous improvement

The score is a starting point.

Most methodologies stop at measurement. This one is built to run in a loop.

Observe
Measure
Improve
Validate
Research
Repeat

See where your store stands

Free AI Commerce Score across all 8 layers.

Find out exactly what AI agents see when they evaluate your store against every measurement layer described above.

Free · No signup · Results in 10 seconds