What is an agent readiness score, and how do you check yours?

An agent readiness score is a number from 0 to 100 that rates how well AI agents can fetch, understand and act on your website. AI agents here means the crawlers and assistants behind ChatGPT, Claude, Perplexity and Gemini, plus the browser agents that shop and book on someone's behalf.

It is a technical measurement, not a visibility measurement. A high score means agents can read your site without guessing. It does not mean they recommend you. That second question needs a different instrument, which we come back to at the end.

Why a score at all

Search engines taught site owners to think about one reader: Googlebot. It renders JavaScript, follows every link and comes back tomorrow.

AI agents behave differently. Most of them fetch a page once, skip JavaScript, and work with whatever HTML arrives. If your robots.txt or your CDN blocks them, they do not throw an error. They answer anyway, with less information about you. We wrote about that failure mode in When an AI agent cannot read your page, it still gives an answer about you.

A score turns a dozen scattered signals into one number you can track, compare with competitors and hand to a developer with a list of fixes. Google Lighthouse did the same for page speed.

What goes into an agent readiness score

Every scanner weights things differently. The AgentReady Check measures nine areas, grouped into four questions an agent asks in order.

Retrievable: can an agent fetch and read the page? (51 points)

AI crawler access (18 points). Does robots.txt allow GPTBot, ClaudeBot and PerplexityBot, and does your homepage return the same status code to them as to a browser? Firewalls and bot protection often block crawlers that robots.txt allows. Fix guide: How to allow GPTBot and other AI crawlers in robots.txt.
JavaScript dependency (14 points). How much readable text is in the HTML before any script runs? Under 1,500 characters is critical. Fix guide: How to make a JavaScript site readable to AI crawlers.
Agent basics (10 points). sitemap.xml, canonical URLs, a findable contact and legal page, and a homepage that responds in under 3 seconds.
Search index verification (9 points). Public Google Search Console and Bing Webmaster verification, and a sitemap declared in robots.txt that actually resolves. Many AI search products lean on these indexes.

Explainable: can an agent extract accurate facts? (37 points)

llms.txt (14 points). A plain-text overview of your offering at /llms.txt, served as text and not as an HTML soft 404. Fix guide: What is llms.txt, and is it worth adding?
Structured data (14 points). JSON-LD with Organization, Product, Offer or FAQPage, and prices stored as numbers rather than text.
Feed and product data (9 points). For shops: name, price, currency, availability, identifier and image on every product, plus machine-readable return and shipping terms. Background: How AI agents choose between products.

Actionable: can an agent use your offering? (7 points)

Agentic buyability (7 points). Signs of a checkout or API an agent can call. More in Agentic checkout in 2026.

Distributed: are you in the catalogues agents shop from? (5 points)

Catalog prerequisites (5 points). Whether your product data meets the preconditions for merchant catalogues.

How the number is calculated

The score is not a plain sum. It is the share of points that actually apply to your domain:

score = round(100 × points earned ÷ points applicable)

A blog has no checkout and no product catalogue, so those areas drop out instead of counting as zero. Signals that cannot be seen from outside, such as Search Console verification by file upload, count as half rather than zero. The full logic is on How the check works.

What a good score looks like

0 to 39 (red): agents get no reliable picture of your site. Start with crawler access and server-rendered text.
40 to 69 (yellow): the foundations are there, but agents have to piece the important facts together themselves. Usually llms.txt and structured data are missing.
70 to 100 (green): you pass most applicable technical checks. From here, what is written about you matters more than your markup.

For live numbers across every domain we have scanned, see the live market data.

How to check your agent readiness score

Open the free AgentReady Check and enter your domain. No login.
Wait 5 to 30 seconds while it fetches your homepage, robots.txt, llms.txt, sitemap and DNS records, and tests your homepage with real AI crawler user agents.
Read the result: a score, a traffic light, and one concrete next step per finding, each linked to a fix guide. Three of the nine checks show right away; the other six appear after you enter an email address.

The same scan runs from code. The API returns all nine checks as JSON, and AI assistants that support MCP can call it as the scan_domain tool.

Agent readiness score vs. AI visibility

A perfect technical score is necessary, not sufficient. ChatGPT can read every word of your site and still recommend a competitor, because it weighs reviews, comparisons and how clearly your prices and differences are stated.

So measure both:

Agent readiness: can agents read you? That is the free check.
AI visibility: do agents recommend you when a buyer asks? That is what AgentReady Monitor tracks each month, with the answers models give for real buying questions in your category.

FAQ

Is an agent readiness score the same as an SEO score?

No. They overlap on sitemaps, canonical URLs and speed, but SEO tools assume a crawler that renders JavaScript and ranks links. An agent readiness score also checks AI crawler access, llms.txt, machine-readable prices and agent-callable actions.

Does a higher score get me cited in ChatGPT?

It removes the technical reasons agents skip you. Whether they cite or recommend you depends on your content and on what others write about you.

How often should I re-check?

After every relaunch, CDN or firewall change, and whenever you change robots.txt. Bot protection updates can block AI crawlers without anyone on your team noticing.