How to Audit Your Website for AI Readiness (Free Method)

An AI-readiness audit answers one question: if an assistant or agent retrieves your site, what can it actually do with it? We split that into three weighted categories. Discovery (40%) covers whether AI crawlers, GPTBot, OAI-SearchBot, ClaudeBot, Google-Extended, PerplexityBot, can fetch your pages at all, plus your sitemap, llms.txt, canonical URLs, and rendering without JavaScript.

Data Quality (40%) covers what they find when they get there: title tags, meta descriptions, one clear H1, JSON-LD for Organization, Service, and FAQPage, image alt text, internal links, published pricing, and whether the content has real depth. Actionability (20%) covers what happens next: discoverable docs, FAQ coverage, MCP signals, a clear conversion path, and pre-authorized agent actions that route sensitive steps to human approval.

The weights are our judgment, not an industry standard: 40/40/20 reflects the reality that most sites fail on legibility long before they fail on tooling. Fix Discovery first, because a blocked crawler never gets to see your Data Quality work.

Why audit at all, rather than keep a checklist in a doc? Because the failure mode is invisible from the inside. Your site looks fine in a browser, you are logged in, scripts run, and you already know the answers. An audit asks the other questions: what does a sessionless crawler see, what can a model reconstruct from fragments, and which of your facts are simply never published anywhere a machine can reach.

You do not need our tool to get a baseline. This is the hand-run version, seven steps, no login, no install, and every result is something you can verify yourself in a terminal or a browser tab.

Two steps deserve emphasis. The robots.txt read tells you which systems are welcome at all; the no-JS test tells you whether the welcome matters, since content that only appears after scripts run may never reach a crawler. And the assistant test is the only step that measures outputs rather than inputs, it shows what a model already believes about your company, which is the thing you are ultimately trying to correct.

The manual pass takes an hour the first time and still misses things. The free AI Agentship audit runs about 40 checks in roughly 30 seconds: enter your domain, and it scores Discovery, Data Quality, and Actionability, then returns a prioritized fix list with copy-ready code, the exact robots.txt rules, JSON-LD blocks, and meta tags to paste in, ordered by impact.

Every run produces a shareable report URL, so you can send the findings to a developer or a partner without exporting anything. The summary is free; the full report unlocks with a name and email. If you want the reasoning behind each individual check rather than a list, the agent-ready checklist is the same audit written out by hand, 27 checks at a time.

Here is the honest part: the score measures legibility. It tells you how easily AI systems can fetch, parse, and act on your site. It does not predict whether an assistant will cite you, whether you will rank, or whether traffic, leads, or revenue will move. No audit, ours or anyone else's, can promise that, and we do not sell that promise.

Think of the score like a lighthouse reading: it tells you whether the lamp is lit and the glass is clean, not how many ships will come. What it does give you is a defensible work plan, highest-impact fixes first, with the code ready, instead of a vague feeling that you should "do something about AI" before the next quarter.

It is also not a grade you pass once. Scores drift: a plugin adds a redirect, a redesign ships with noindex, a staging rule survives launch. Re-run the audit after any meaningful publish and you will catch it in 30 seconds rather than in a quarter of missing visibility.

Most audits land in the same place: a short priority-1 list, robots.txt rules, missing schema, content trapped behind JavaScript, unpublished pricing, and a longer tail of polish. Work the priority-1 items with the copy-ready snippets, then re-run the audit to confirm the score moved. That loop, repeated quarterly, is the whole method.

If you would rather have it done for you, pricing shows the range: the online audit is free, builds start at $2,500, and a full multi-page build runs $15,000-$25,000. Start with the free run either way, it is 30 seconds and it tells you whether the rest of this article is even your current problem.

One warning about first runs: resist the urge to redesign. The most common outcome is that a handful of small, unglamorous fixes move the number, one robots.txt block removed, one leftover noindex dropped, one pricing paragraph added, three JSON-LD blocks pasted in. The audit exists to stop you from paying for the big fix before you have tried the small ones.