Arrow 
Fifteen checks test whether an AI agent can reach your pages, understand your offer, complete the next step and verify your claims. Fix access first, then price and next-step gaps, then contradictions and proof. One afternoon is enough for a first review.
How do you know if your website is agent-ready?
Test whether an AI agent can reach your pages, understand your offer, complete the next step and verify your claims. The fifteen checks below cover those four areas. You can run them in an afternoon with a browser, an AI assistant and a spreadsheet. Mark each check pass, partial or fail, and fix the failures that block a customer decision first.
Access: can agents reach the pages?
| # | Check | Pass when |
|---|---|---|
| 1 | Crawler policy | robots.txt does not block the AI search and agent user agents you want to serve |
| 2 | Content in HTML | Key facts appear in the page source, not only after scripts run |
| 3 | No wall before facts | Prices, services and hours are not behind a login, cookie wall or pop-up |
| 4 | Fast, stable pages | Pages load reliably without timeouts or endless redirects |
Start with the AI crawler guide for check 1. OpenAI documents its crawlers and user agents in its bot documentation. Blocking a training crawler is a separate decision from blocking a search or agent user agent.
Clarity: can agents understand the offer?
| # | Check | Pass when |
|---|---|---|
| 5 | One-sentence offer | Each service page states what it is in the first paragraph |
| 6 | Audience and limits | The page says who it is for and who it is not for |
| 7 | Price or logic | A price, starting price or pricing rule is written in text |
| 8 | Place and time | Service area, address and hours are explicit and current |
| 9 | Structured data | Schema markup matches the visible facts |
Check 9 is covered in schema markup for AI agents. The machine-readable content checklist adds detail on formats.
Action: can agents complete the next step?
| # | Check | Pass when |
|---|---|---|
| 10 | Labelled forms | Every field has a visible text label |
| 11 | Requirements upfront | Documents, deposits or eligibility rules are stated before the form |
| 12 | Text confirmation | A submission returns a clear confirmation in text, not only an animation |
Booking and intake design is covered in booking and intake flows an AI agent can complete. CAPTCHAs and fraud checks are legitimate; the aim is not to remove protection but to avoid hiding the steps themselves.
Trust: can agents verify your claims?
| # | Check | Pass when |
|---|---|---|
| 13 | Dated proof | Case studies and reviews carry dates and attributable names |
| 14 | Consistent profiles | Business profiles and directories match the website |
| 15 | Ownership | One person or team is responsible for keeping each fact current |
Check 15 is the one most teams skip. Facts drift when nobody owns them. See who owns AI visibility for a simple governance model.
How should you run the test?
Use one real delegated task per main service. Ask an AI assistant with browsing or agent features to complete it on your site and note where it stops or hedges. Then go through the fifteen checks manually on the same pages.
Record the date, the assistant, the task wording and the outcome. That record lets you rerun the same test after a fix and see whether the agent's behaviour changed. One run is an observation, not a benchmark.
What should you fix first?
Fix in this order: anything that blocks access, then missing price and next-step facts, then contradictions between your site and your profiles, then structured data and proof. Access failures hide everything else, while proof improvements matter most once the agent can already understand and act.
Return to the agent-ready business guide for the strategy behind the checklist, or request a free GEO audit for an outside review.
Sources and editorial scope
Reviewed October 1, 2026. Platform-specific statements link to the documentation beside the claim. The workflows and examples are Arrow AI editorial guidance. Agent features change quickly and vary by plan and country; check the linked documentation before acting. Fictional scenarios illustrate a method; they are not customer results, measured demand or evidence of ranking gains.
Part of Agent-Ready Business. Explore Arrow GEO or request a free audit to see which facts an AI agent can already find about your business.