Agent Readiness
What Fond's Agent Readiness checks measure, how the passing count is worked out, what each check status means, how often checks run, and how to fix a failing one.
Agent Readiness is a list of checks on whether an AI assistant or agent can find your site, read it and act on it. Find it under Analytics → Agent Readiness, or on the AI Readiness tile of the website overview. To improve it, give the fix of a failing check to the person who edits your site, then click Re-check.
Read the score
The headline reads X of Y checks passing. X is the checks that passed, and Y is the checks that passed or failed. Every check counts the same, whatever its severity. There is no letter grade or percentage.
Fond leaves two kinds of result out of both numbers:
- a check that Fond could not complete, because it tells you nothing about your site
- a check that does not apply to the kind of site you chose, or that you turned off
The line under the headline names the checks that Fond could not complete and how many you turned off. Checks outside your kind of site say so on their rows. If no check can be scored, the page says so instead of 0 of 0. Each category shows the same count for its own checks.
What each status means
| Status | What it means | Counts toward the score |
|---|---|---|
| Pass (green tick) | Fond looked and found what the check asks for. | Yes |
| Fail (red cross) | Fond looked and it is not there. | Yes |
| Not checked | Fond could not complete the check. Your site did not respond or blocked Fond's crawler, Fond had a fault, or Fond cannot read this data. The row gives the cause when Fond knows it. This is not a failure to fix. | No |
| Doesn't apply | The check is outside the kind of site you chose, or you turned it off. Expand the row to see which. | No |
Fix a failing check
Expand a failing row to see:
- what passing looks like, under Goal
- what Fond found, under Issue
- the fix in plain steps, under How to implement
- links to the relevant standard or platform documentation, under Resources
- the requests that Fond made and what came back, under Audit details, where Fond recorded them
Click Copy prompt to copy the check, goal, issue, fix and links as a prompt for a coding agent or developer. Copy all fixes, at the top of the page, does the same for every failing check.
Severity and effort
Every failing check carries a severity label (Critical, High, Medium or Low) for how much the gap is likely to cost you with AI assistants. It also carries an effort label (Trivial, Moderate or Involved) for about how much work the fix takes.
In each category, failures come first, by severity and then by least effort. The category with your most severe failure opens automatically. Neither label changes the score.
The check categories
AI assistants learn about your site from crawlers that read your pages ahead of time, and from agents that visit while they answer someone. Things that a human does not notice can stop either one. Fond groups the checks in the order that an assistant meets your site:
- Bot access control checks whether your robots.txt lets the main AI crawlers,
such as GPTBot, ClaudeBot and PerplexityBot, crawl your site from its root.
Fond reads the rules for
/only, so a rule that blocks only/products/does not fail these checks. - Discoverability checks for a sitemap, and an llms.txt file with useful structure and content.
- Content accessibility checks whether your homepage content is in the HTML itself, not drawn only by JavaScript. It also checks that a missing page returns a real 404.
- Entity checks whether assistants can tell that your brand is one specific company. It looks for a Wikidata item, Wikipedia coverage, and your official profiles linked from your homepage.
- Commerce, for online stores, checks whether an agent can read your products, prices, availability, return policy and shipping details as structured data.
- Protocol discovery covers newer standards that let agents discover what your site offers and complete a checkout. These checks carry an Emerging label because few sites use them yet, but they count toward your score.
The page lists a check, or a group of checks, only after it has a result.
Choose which checks apply
Click Configure to tell Fond what kind of site this is.
- Choose Online store (the default) if you sell products directly on this site. Every check applies.
- Choose Content site if you do not sell products here. The commerce checks and a few other store-only checks show as Doesn't apply.
You can also turn off individual checks. A check that you turn off shows as Doesn't apply and leaves the score. Fond keeps checking it, so if you turn it back on, it shows its latest result.
When checks run
Every six hours, Fond queues any site whose results are more than five days old, or that it has not checked yet. So Fond checks a site that it can read about every five days.
If Fond cannot read your site, it tries again every six hours. After three failures in a row, it tries once a day until it succeeds.
Results arrive in groups, so a new site can show some categories before others. The line under the headline says how many groups have run. The date under it is when Fond checked the oldest result on the page.
To check sooner, for example after a fix, click Re-check. You can do this once every ten minutes.
If another website shares your domain
Fond stores check results once for each domain. This affects you only when
another website in Fond, in any account, is on your domain, such as
shop.example.com beside example.com.
The two websites then share one set of results, checked from the website that the schedule reached first. So the homepage, product page and robots.txt checks can describe the other address. The Storefront API check stays your own, but the schedule can skip its refresh. Re-check checks your own address and refreshes that result. A later scheduled check can replace the shared results again.
What Agent Readiness does not measure
- Whether assistants mention you. That is AI visibility, on the Visibility page.
- Whether your firewall lets AI crawlers in. A firewall or bot filter can block a crawler that robots.txt allows, and Fond cannot reliably test that. If your site turns away Fond's crawler, the affected checks show as Not checked. This shows how your site treats Fond, not an AI crawler.
- Every page. The product data checks read one product page, and most other page checks read your homepage.
- The weight of each gap. Use severity to decide what to fix first.
Tracked competitors and brand merges
Which competitors you can track and what tracking changes, how Fond decides two names are one brand, and how to merge, rename or ungroup brands on the Competitors page.
Site Audit
How Fond's site audit is made, how to read a finding and its priority, what the coverage notices mean, and what approving or declining a finding does.