How AgentReady scores your site

7 dimensions. 100 points. Each one reflects a specific step in the AI agent buying journey — from initial discovery through to autonomous transaction.

Version 1.0 · Updated May 2026 · Scores recalculated monthly

The agent buying journey

When a buyer delegates a purchase to an AI agent, it executes this sequence. Your score reflects how far it gets before it fails — or succeeds.

🔍
Step 1

Discovery

User asks the agent: "Find me the best waterproof dog bed under $100 with free returns."

Dimensions tested

AI VisibilityMachine Readable

Agent identifies candidate brands from training data and live search.

⚖️
Step 2

Evaluation

Agent crawls shortlisted sites, extracts product specs, pricing, and trust signals.

Dimensions tested

Content ExtractabilityMachine ReadableEntity Consistency

Agent scores each brand on completeness of extractable evidence.

📋
Step 3

Recommendation

Agent builds a shortlist ranked by confidence. It needs proof to justify its top pick.

Dimensions tested

Competitive ReadinessAI Visibility

Agent presents a ranked recommendation with supporting evidence to the buyer.

Step 4

Transaction

Buyer approves. Agent executes the purchase — or stalls because it can't find a clear path.

Dimensions tested

ActionabilityAgent Governance

Purchase completes — or agent returns to the buyer with a friction point.

🔒 Human Approval Gate

Before any agent completes a purchase, the buyer confirms. Your score predicts how likely they are to approve: high-scoring sites give agents complete, trustworthy evidence — buyers approve quickly. Low-scoring sites produce vague recommendations — buyers hesitate or reject.

Score bands and approval rates

Each tier represents a distinct agent capability level, mapped to a measurable human approval rate.

91100
AgentReady

You win in autonomous commerce.

85–95%+

approval rate

7690
Agent Preferred

Agents prefer you — buyers approve.

65–85%

approval rate

6175
Agent Recommendable

You're in the recommendation — not always the winner.

45–65%

approval rate

4160
Agent Understandable

You're understood, but not recommended.

20–45%

approval rate

2140
Agent Weak

You're found but not understood.

5–20%

approval rate

020
Agent Invisible

You don't exist in AI commerce.

< 5%

approval rate

The 7 dimensions in detail

Each dimension measures a distinct capability in the agent buying journey, with weighted sub-signals.

1

Machine Readable

/18 pts

Can an AI agent parse your products without guessing?

Machine readability is the foundation. Without structured data, agents treat your site as an unstructured blob of text — they might guess at prices and product names, but they won't commit that guess to a recommendation. JSON-LD Product schema gives agents exact pricing, availability, condition, and brand in a format they can parse in milliseconds.

Sub-signals tested

High

JSON-LD Product schema present

Does every product page emit valid schema.org/Product markup?

High

Price & availability accuracy

Are price, priceCurrency, and availability fields populated and current?

Medium

Breadcrumb + site navigation schema

Can agents map your catalogue hierarchy from BreadcrumbList markup?

Medium

Organization / LocalBusiness schema

Is your business identity declared with schema at the site root?

Low

FAQ and HowTo schema (where applicable)

Do support pages and product guides use structured markup?

High

Schema validation (no errors)

Does schema.org validation return clean output with no critical errors?

Quick fix: Install a schema plugin or add JSON-LD Product blocks to your product template. Validate at schema.org/validator.

2

Actionability

/18 pts

Can an agent actually buy — not just browse?

Reading your site is step one. Acting on it is step two. Actionability measures whether an agent can find and execute a purchase path: add to cart, proceed to checkout, complete a transaction. Without explicit action pathways, agents stall at the research phase and never convert. /llms.txt is the new robots.txt — it tells agents what they're allowed to do on your site.

Sub-signals tested

High

/llms.txt published

Have you declared agent permissions, allowed actions, and contact info at /llms.txt?

High

PotentialAction schema on product pages

Do product pages declare a BuyAction or OrderAction with a target URL?

High

Cart/checkout API or Headless endpoint

Is there a programmatic path to add items and initiate checkout?

Medium

Contact and support action paths

Are ContactAction or CommunicateAction schemas declared for agent queries?

Medium

Search action schema

Does your site declare a SearchAction so agents can query your catalogue?

Low

Agent-accessible order status

Can agents check order status on behalf of customers?

Quick fix: Publish /llms.txt (template at llmstxt.org). Add PotentialAction schema to product pages with a target checkout URL.

3

Entity Consistency

/18 pts

Does every source agree on who you are?

AI agents are trained on the web. If your business name appears differently across sources, agents see multiple possible entities and lower confidence scores across all of them. Entity consistency is the trust layer — it tells agents that your site, your Google Business profile, your LinkedIn, and your schema declarations all refer to the same verified business.

Sub-signals tested

High

Business name consistency (site vs Google vs LinkedIn)

Does your exact brand name match across all major indexed sources?

High

Phone number NAP consistency

Is the same phone number on your site, Google Business, and directories?

Medium

Address consistency

For businesses with a physical location: does your address match exactly?

Medium

Logo URL stability

Does your schema Organization entity reference a stable, crawlable logo URL?

Medium

Social profile cross-links

Are your sameAs links in Organization schema pointing to active, verified profiles?

Low

Review profile consistency

Does your brand name on Trustpilot, Google Reviews, and G2 match exactly?

Quick fix: Audit your name, phone, and address across Google Business, LinkedIn, Yelp, and Trustpilot. Update your Organization schema sameAs array to link them all.

4

Content Extractability

/14 pts

Can agents read what you've written?

JavaScript-rendered content is invisible to most AI crawlers. If your product descriptions, specs, and pricing are injected by React or Vue after page load, agents see only an empty shell. Similarly, an overrestrictive robots.txt blocks agents from the pages that matter most. Content extractability measures whether your actual words are available to agents at crawl time.

Sub-signals tested

High

robots.txt allows AI crawlers

Are GPTBot, ClaudeBot, anthropic-ai, and PerplexityBot allowed on product pages?

High

Server-side or static rendered content

Are product names, prices, and specs in the raw HTML — not injected by JS?

Medium

Content density (text/HTML ratio)

Is your page content substantive, or mostly navigation and boilerplate?

Medium

Image alt text completeness

Do product images have descriptive alt text agents can use as product context?

High

No aggressive bot blocking (Cloudflare/CAPTCHA on crawl)

Are AI crawlers reaching your pages, or being blocked by WAF rules?

Low

Sitemap coverage

Does your sitemap.xml include all product pages agents need to discover?

Quick fix: Allow GPTBot and ClaudeBot in robots.txt. Ensure product data is server-rendered. Check your CDN/WAF isn't blocking AI user agents.

5

AI Visibility

/14 pts

Do the AIs already know you exist?

AI training data is not just the web — it's the web as it was indexed, cited, and discussed in high-quality sources. If ChatGPT, Claude, and Gemini have never seen your brand mentioned in a press article or review site, you're not in their world model. AI Visibility measures your pre-existing presence in the sources AI models draw on when generating recommendations.

Sub-signals tested

High

Mentioned in AI model outputs (Claude, ChatGPT, Gemini)

Does asking AI assistants about your category surface your brand?

High

Press / media citations

Has your brand been mentioned in news outlets or trade publications that AI models train on?

Medium

Review platform presence (Trustpilot, G2, Capterra)

Do major review aggregators have an active listing for your brand?

Medium

Industry directory listings

Are you listed in category-relevant directories that AI models cite?

Low

Wikipedia or Wikidata presence

For larger brands: does your entity appear in Wikidata with accurate claims?

High

Perplexity / AI search citation rate

When Perplexity answers category questions, do you appear in source citations?

Quick fix: Build citations: contribute to industry publications, get listed on review platforms, and earn press coverage. Create content that earns links from AI-indexed sources.

6

Competitive Readiness

/10 pts

Can agents justify recommending you over rivals?

An agent doesn't just need to find and understand you — it needs to argue for you. When a buyer asks which brand to buy, an agent runs a comparison. Competitive readiness measures whether you've given agents the proof signals they need: AggregateRating data, a differentiator statement, and explicit comparison content. Without these, agents default to recommending the best-documented competitor.

Sub-signals tested

High

AggregateRating schema with count and value

Do your product pages declare a valid AggregateRating with reviewCount and ratingValue?

High

Clear differentiator statement

Is there an explicit, crawlable statement of what makes you different from competitors?

Medium

Comparison page or comparison content

Do you have a page comparing your products to named competitors?

Medium

Trust signals (certifications, guarantees, press logos)

Are trust badges and proof points machine-readable and not just images?

Low

Price positioning signal

Is your pricing tier (budget / mid / premium) clear to an agent parsing your site?

Quick fix: Add AggregateRating schema to product pages. Write a clear 'Why us?' page with crawlable comparison content. Ensure trust signals are text-based, not image-only.

7

Agent Governance

/8 pts

Have you told agents what they're allowed to do?

As agents move from research to transaction, they need explicit permission from site owners. /AGENTS.md is the emerging standard: a document declaring which agents are permitted to act on your site, what actions they can take, and how disputes are handled. Without it, well-designed agents treat your site as a no-autonomous-transaction zone — they'll research but won't buy.

Sub-signals tested

High

/AGENTS.md published and valid

Does your site publish an /AGENTS.md declaring agent permissions and allowed actions?

High

Agent purchase workflow documented

Is there a documented process for how agents initiate, confirm, and complete purchases?

Medium

Data access policy for agents

Have you declared what personal data agents can access or store on behalf of buyers?

Medium

Dispute and cancellation process

Is there a machine-readable process for agents to cancel or modify an order?

Low

Agent authentication method

Is there an OAuth or API-key based method for agent identity verification?

Quick fix: Publish /AGENTS.md at your site root. See agents.do or llmstxt.org for templates. Document which agent actions are permitted and how purchases are authorised.

How scoring works

Each domain is crawled using a headless browser that captures raw HTML, JavaScript-rendered content, structured data, robots.txt, and accessibility files including /llms.txt and /AGENTS.md.

Technical signals (schema presence, robots rules, token counts) are scored deterministically. Qualitative dimensions (content quality, actionability depth, competitive differentiation) are assessed using Claude AI with a structured rubric that returns consistent, auditable JSON scores.

Entity consistency is checked by comparing business facts scraped from Google's knowledge panel, LinkedIn public pages, and Yelp against the domain's own website content and schema declarations.

AI visibility is tested by running a standard battery of prompts against live Claude and ChatGPT instances, recording mention rate, citation rate, and whether the brand is recommended or excluded.

Human approval likelihood is a directional model — it reflects the relationship between score completeness, evidence quality, and an agent's ability to build a verifiable recommendation buyers trust. It is not a guaranteed conversion rate.

See where your site falls across all 7 dimensions.

Score your site free →