What does Peaje's radar measure and what does my score mean?
Where the score from Peaje's radar comes from, how much each part weighs, what each grade from A+ to F means, how the "with Peaje" number is calculated and how to read the rest of the result.
Your score, from 0 to 100, says how ready your site is for an AI agent to find you, use you and pay you. It is calculated by Ora, an independent scanner that checks your site in four layers, and your grade, from A+ to F, comes from there too. The "with Peaje" number is an estimate of how much you would gain by fixing what the kit covers, and it is not a promise.
Updated: November 2026
According to Ora's methodology, the 100 points are split like this: 20 for being found, 30 for letting agents in, 40 for agents being able to use your site and 10 for being able to pay you.
What does each part check?
The radar groups Ora's four layers into three questions:
| On the radar | Ora layers | Points | What it checks, for example |
|---|---|---|---|
| Can they find you? | Discovery | 20 | What your robots.txt tells AI bots, whether you show up on top when people search your brand and whether your MCP server is listed in directories |
| Can they use you? | Access and Usability | 70 | llms.txt, sitemap, structured data, readable prices, pages that work without JavaScript and an MCP server that works well |
| Can they buy from you? | Payments | 10 | Whether you accept any of the protocols agents pay with: ACP, UCP, AP2, x402 or MPP |
Being usable is worth 70 of the 100 points, so fixes in that part usually move your grade the most.
What does my grade mean?
It is Ora's scale:
| Grade | Score | What it means |
|---|---|---|
| A+ | 95 to 100 | Leading |
| A | 86 to 94 | Agent-ready |
| B | 70 to 85 | Competitive |
| C | 48 to 69 | Needs work |
| D | 28 to 47 | At risk |
| F | 0 to 27 | Agents can barely use it |
In the example on the radar page, a site at 46 (D) could reach about 74 (B) with the kit.
What does "agents understand 14 of 29 things" mean?
It is the sentence that opens your result. The first number is the checks your site passes, and the second is the checks that apply to your site and could be measured. That is why the total changes from one site to another, and anything that could not be measured that day goes under "Could not be measured" and counts neither for nor against you.
How is the "with Peaje" number calculated?
It adds to today's score the points of every failing check that Peaje's kit can fix, like llms.txt, JSON-LD or the MCP server, without going over 100. What the kit does not cover, like the text on your pages or your checkout, is left out of that sum. If your site already passes everything the kit covers, the radar tells you.
How do I read the rest of the result?
- What is missing: each fix with the points it adds (+6, for example), sorted with the easiest first and, within that, by what raises your score the most.
- What AI answers: the radar asks ChatGPT and Claude up to 3 questions from someone looking for what you offer, without naming your brand, and shows you whether they cite you, name you without a link or leave you out, and who they cite instead. That part is kept for 7 days.
- What you already do well: the checks you pass.
You can share the result with a link, for example with whoever runs your website. If the measurement is more than 7 days old, the radar warns you that your site may have changed.
Does my score say whether ChatGPT recommends me?
Not entirely. The score measures whether an agent can read and use your site, which is the base, and the questions to ChatGPT and Claude are a snapshot of that day. Getting cited also depends on your content and on what other sites say about you, and answers change from one day to the next. The Pro agent tracks that, with 20 questions every week (how do I know if they cite me).
How do I raise my score?
Start with what the radar puts at the top of "What is missing". Peaje's kit is free and covers a good part of it: you get it by registering your site at usepeaje.com/nuevo and you install it from your dashboard, with a text for your AI coding tool or with a command. Then you measure again on the radar, which has no limit on measurements beyond 6 per hour.