What is an agentic score, and how is one calculated?
An agentic score, also called an agent readiness score, is a number that rates how well an AI agent can use a website: find what it needs, read it correctly and complete a task such as a purchase. The Axidort Score is an agentic score from 0 to 100 built from seven measures, with a grade from A to F.
By the Axidort team · Updated
What an agentic score tells you
A search ranking tells you whether people can find your pages. An agentic score tells you whether software acting for a person can finish the job once it arrives. That software might be a shopping assistant comparing three suppliers, a browser agent filling a cart, or a procurement agent checking contract prices.
There is no single standard score. Tools that publish one fall into two families, and the number means something different in each.
| Kind of score | What it checks | What it cannot tell you |
|---|---|---|
| Standards score | Whether files and protocols exist: robots.txt rules for AI bots, a sitemap, Markdown responses, an MCP server card, OAuth discovery. | Whether an agent can actually complete a task on the site. |
| Behavioural score | What happens when a real agent attempts a journey: search, open a product, read price and stock, reach checkout. | Whether you support a specific emerging protocol in full. |
The Axidort Score is mostly behavioural: what the agent manages to do carries 90% of the score and file-based signals carry 10%. A site can publish every recommended file and still stop an agent at a bot challenge, so the two kinds of score are complements, not substitutes. See agent readiness checkers compared.
How the Axidort Score is calculated
Two agents visit the site. A browser agent searches for a product, opens it, reads price and stock and tries to reach checkout. A second agent reads the same page without a browser, the way an AI assistant fetches a URL. Nothing is ordered. The result is scored on seven measures.
| Measure | Weight | What earns the points |
|---|---|---|
| Product discoverability | 20% | A product can be found through search or navigation in a few steps. |
| Checkout reachability | 20% | The agent can get from a product page to the purchase step. |
| Product data quality | 15% | Name, price, availability and SKU or GTIN are readable without running JavaScript. |
| Pricing visibility | 15% | A clear price with currency and unit. |
| Stock availability | 10% | Availability stated as text and in structured data. |
| Agent accessibility | 10% | No bot challenge, rejected request or pop-up wall in the way. |
| Static signals | 10% | robots.txt rules for AI agents, sitemap, llms.txt, structured data, feeds, API or MCP signals. |
Measures that do not apply to a site are left out and their weight moves to the rest, so an information site is not marked down for having no cart. The full rules are in the methodology.
What is a good agentic score?
| Grade | Score | Meaning |
|---|---|---|
| A | 85 to 100 | Agent-ready |
| B | 70 to 84 | Mostly ready |
| C | 50 to 69 | Partially ready |
| D | 30 to 49 | Agent-hostile |
| F | 0 to 29 | Invisible to agents |
Most sites are not close. Across the 50 leading commerce sites in the Axidort Index, the median Axidort Score is 11 out of 100. 7 sites reach grade A or B and 36 are graded F (scanned October 1, 2026). A score above 70 already puts a site in a small group.
How to raise an agentic score
- Remove the blocker first. A bot challenge or a rejected add-to-cart request caps everything else. See why AI agents fail at checkout.
- Put the facts in the HTML. Price, currency, unit and stock as text in the server response, mirrored in structured data.
- Make the journey short and standard. Working search, real buttons and links, guest checkout where you can. See how to make a website agent friendly.
- Add the files. robots.txt rules for AI agents, a sitemap and llms.txt.
- Rescan. A score is a measurement at a point in time. Scan again after each change and track the movement.
Frequently asked questions
01Is an agentic score the same as an agent readiness score?
Yes. The two terms are used for the same idea: a rating of how well AI agents can use a website. Different tools calculate it differently, so compare scores from the same tool only.
02What is the average agentic score?
In the Axidort Index of 50 leading commerce sites the median Axidort Score is 11 out of 100, and 72% of the sites are graded F. The figures are from scans on October 1, 2026 and change with each rescan.
03Can I get an agentic score for free?
Yes. The Axidort scan is free during the beta. You enter a domain, sign in with Google to view the report, and get a score, seven component scores and a grade.