How AgentScore works: what's behind the 100 points
AgentScore gives your website one number out of 100 for how well AI agents can get in, read your pages and find their way around. A single number is only useful if you trust it, so here is exactly how it's worked out: the four categories, how each check earns points, which pages the scanner visits, and what it doesn't score yet.
Four categories, 100 points
Every check belongs to one of four categories. Each category has a fixed number of points, set by how much it matters to an AI agent trying to do something for a customer.
| Category | Points | The question it answers |
|---|---|---|
| Access | 25 | Can AI agents get in, past robots.txt, bot protection and challenges? |
| Readability | 25 | Can they understand your pages without guessing? |
| Navigability | 20 | Can they find and use the buttons, links and forms they need? |
| Task completion | 30 | Can an AI agent actually finish a real task, such as reaching checkout? |
Points per category
- Access 25
- Readability 25
- Navigability 20
- Task completion (free scan, when an agent runs) 30
The background to these categories is in our guide to agent readiness. This post is about the mechanics.
How a check earns points
Each check ends in one of five results:
- Pass earns the check's full weight.
- Warning earns half. Something works, but agents will stumble on it.
- Fail earns nothing.
- Couldn't test earns a quarter. Your site stopped the check from running: a key page errored, was blocked or wouldn't load in a browser. Agents meet the same wall.
- Not tested is left out of the sum entirely, because the check doesn't apply to your site. It never counts against you.
Within a category, checks carry different weights, because some problems stop an agent cold and others only slow it down. A category's score is its points multiplied by the share of weight you earned among the checks that could be tested. If Readability's tested checks are worth 17 in total and you earn 13.5 of them, you get 25 × 13.5 ÷ 17, or 19.9 out of 25.
Checks on the basics of a page (titles and headings, content without JavaScript, image descriptions, button names, form labels, landmarks and real buttons) run on every page the scan looks at, not just your home page. Each one scores your typical page, with your worst page pulling it down: 70% comes from the median page and 30% from the worst. A tidy home page can't hide problems everywhere else. Product data and prices are checked on every product page the scan finds.
The overall score adds up the categories that could be tested and scales the result to 100. So if Task completion wasn't tested and you scored 20 for Access, 17 for Readability and 11 for Navigability, your AgentScore is 48 out of a possible 70, which is 69.
- 48 points earned
- 70 points that could be tested
- 69 AgentScore, scaled to 100
Most checks are rule-based rather than judged by an AI model, so the same site gets the same score from one scan to the next. When it moves, something on the site changed.
What the bands mean
| Score | Band | What we tell you |
|---|---|---|
| 80 and up | Good | AI agents can use this site well. |
| 60 to 79 | Fair | AI agents can mostly use this site, but some things get in their way. |
| 40 to 59 | Weak | AI agents will struggle on this site. |
| Below 40 | Poor | Most AI agents will fail on this site. |
The one-line verdict at the top of a report also names your biggest issue: the failed check with the highest weight.
Failures that cap the score
Some failures matter so much that nothing else on the site can make up for them, so they cap your score however well the rest does:
- No higher than 59 while AI agents are blocked or challenged: by your bot protection, by a challenge on arrival, on your product, pricing or cart pages, or at the cart and checkout. The same cap applies when the free scan sends an AI agent to do a task and it can't finish.
- No higher than 79 while agents get in but can't rely on what they find: most content only appears after JavaScript runs, prices disagree between the places agents read them, agents can't press the add-to-cart or sign-up button, or checkout needs an account.
When a cap applies, the verdict says which one, and the report shows what you'd score without it. Fix that one finding and the score rises to the number shown.
The checks, by category
Here is what each category looks at today. The weight is how much the check counts inside its category.
Access
- Bot protection lets AI agents through (weight 4). We request your home page as four AI agents and as a normal browser, and compare. If any agent is blocked or challenged while the browser gets through, it fails. If an agent gets less than half the browser's text, it's a warning. We also request it claiming to be Googlebot, as a control: our requests come from our servers, not the AI companies', so if the fake Googlebot is turned away too, your bot protection is checking where requests come from. That's the right setup, so the check is marked not tested. The key pages and checkout checks below work the same way. An agent request that gets no answer within 15 seconds isn't counted as blocked: on the home page and key pages, it's left out of the comparison, and if every request timed out the check is marked not tested.
- robots.txt lets AI assistants and search agents in (3). Blocking one or two assistants or AI search agents is a warning; three or more is a fail. Blocked training crawlers are reported but not scored.
- AI agents can open your product, pricing and cart pages (3).
- AI agents aren't blocked or shown a CAPTCHA at the cart and checkout (3).
- No CAPTCHA or challenge on arrival (2).
- Shoppers can check out without an account (2).
- llms.txt guide for AI (1), Agents can use your site through MCP or WebMCP (1) and Agents can check out through an agentic commerce protocol (1). These are newer standards, so a missing one is a warning, never a fail.
Readability
- Content loads without JavaScript (4): how much of the page's text is in the HTML before scripts run.
- Structured data describes your business and products (3) and Product pages give price and stock in a form agents can read (3).
- Price, stock and name agree everywhere agents read them (3). On each product page we compare the product data, the page as first served, the page after JavaScript runs, and the price tags that social cards and shopping feeds read. A price that disagrees is a fail; a currency, stock or name that disagrees is a warning. Amounts like "free shipping over $50" are ignored, and a variant's price inside the product's price range counts as a match. Prices are read the way shoppers write them: a single dot or comma before groups of three digits separates thousands, so "€1.299" is €1,299 and "$12.990" is 12,990 pesos. When your product data says its price excludes tax, a page price up to 27% higher still counts as a match, since the page may add VAT.
- Clear page title, description, and headings (2), Valid sitemap (2) and Prices are in the page HTML (2).
- Images have text descriptions (1).
Navigability
- Buttons and links have names (4) and Pop-ups and banners can be dismissed by agents (4).
- Agents can find and press your add-to-cart or sign-up button (3) and Cart and checkout fields are labelled for agents (3).
- Form fields are labelled (2), Page has main and navigation landmarks (2), Clickable things are real buttons and links (2) and Key pages are linked from the home page (2).
Weights can change between scoring versions as we learn what trips agents up. Every report records the scoring version it used, so you can compare like with like.
What the scanner visits
AgentScore reads public pages. The only things it presses are your add-to-cart button, once, to see your cart and checkout, and a pop-up's close button if one covers it. It never submits a form, signs up or places an order. In one scan it looks at:
- Your home page, three ways: as a browser sees it before JavaScript runs, rendered in a real browser, and as five AI agents (ChatGPT-User, Claude-User, Perplexity-User, OAI-SearchBot and, as a control, Googlebot).
- Your robots.txt, sitemap and llms.txt, plus the well-known addresses where sites describe an MCP server or an agentic commerce profile.
- Up to three key pages: a product page, a pricing page and the cart. It looks for them the way an agent would, in the links on your home page first, then in your sitemap, then in the product cards on your home page (a picture, a price and a link), then at common addresses such as
/pricingand/cart. Each is fetched as a browser and as an AI assistant, and product and pricing pages are rendered in a browser too. - Your checkout, found from the link on the cart page or at
/checkout, fetched as a browser and as three AI assistants. If the cart is empty and neither shows a form, the scanner adds one item from the product page it checked and loads the cart and checkout again in the same browser session. - Up to six more pages, picked from the links on your home page and spread across your sitemap, with products, categories and other pages such as policies each getting a share. Each is fetched as a browser and rendered in one. Carts, accounts and search results are skipped, and so is any page that won't load in time, so a slow page never stops the scan.
Requests under the scanner's own name are signed so you can verify them. The requests that imitate AI agents are not, on purpose; instead each carries a signed gal_scan tag in its address, so they're told apart from the real agents and left out of your traffic in Ghost Agent Lab. We explain why in Why our scanner sometimes wears other agents' names. How it identifies itself and how to opt out are on the AgentScore bot page.
What isn't scored yet
Task completion, often. When the free scan has an AI agent available, it sends one to try a single task: adding a product to the cart on a store, or finding pricing on a software site. Finishing it in a sensible number of steps passes, finishing slowly is a warning, and failing caps the score at 59. If the agent isn't available, or stopped for a reason that isn't your site's (it ran out of budget, for example), "An AI agent completes a real task" shows as not tested and your score covers Access, Readability and Navigability. Site scans in the Ghost Agent Lab app don't run it; Ghost Agent tests send real AI agents through your journeys there, and their results don't feed into AgentScore yet.
Store-only checks on other sites. Guest checkout, checkout fields, product data and agentic commerce only apply to stores. If we find no products or cart, they're marked not tested.
Anything past the first step of checkout. Many checkouts only open with items in the cart, so the scanner adds one and looks again. It stops there: it never fills in the form, so it can't see later steps such as delivery options or payment. When it can't add an item, for example because a size must be chosen first, those checks say "not tested" rather than guess.
Your choice on training crawlers. Blocking AI training is a business decision, not a readiness problem, so it costs no points.
How to use your score
- Read the verdict and the biggest issue first. It's the highest-weight failure.
- Fix failures before warnings, and Access before everything else: if agents can't get in, nothing else matters.
- Hand each finding to the right person. Most Access problems belong to whoever runs your CDN or bot protection; most Navigability problems belong to developers.
- Rescan after each fix. In the Ghost Agent Lab app you can rescan whenever you like. The free AgentScore page reuses a domain's result for 24 hours, so repeated requests don't hit your site again.
Then go beyond the score: set up tests of your key journeys with AI agents, so you know when a release breaks them.