AI Search
How to Make Your Website Accessible to Perplexity Search
Check whether Perplexity can access your website. Understand its crawlers, safe firewall checks and why accessibility does not guarantee a citation.
If a page is not appearing among Perplexity's sources, there are two separate questions to ask: can the relevant request reach the page, and did the observed answer choose to cite it?
An access check addresses the first question. It cannot guarantee the second. Keep that boundary in place while deciding whether a technical change is needed.
Which Perplexity user agents are relevant to the task?
A user agent is the name software sends with a web request. Perplexity documents its search crawler separately from visits triggered by a user asking to open a page. Ask your developer which kind of request was checked. A user-triggered visit is not proof of automatic discovery or a future citation. Perplexity crawler documentation
Ask the developer to attach the documentation used to the finding. If an article quotes a crawler name, it should explain why that agent matters to the check rather than simply displaying technical terminology.
A practical workflow for the owner and the person making the change.
- Find
Record the page and the actual problem.
- Explain
State why it matters to the customer.
- Assign
Agree who makes the change and what is included.
- Retest
Check the public result before calling it complete.
What should you check in robots and server behavior?
Review the intended rule for the relevant path, then examine how the server actually responds. Keep the URL, response status, date and request context together. A written access rule is only one part of the review. Hosting controls, authentication or other request handling may also affect what is returned, so avoid declaring the whole site accessible from one isolated check.
The technical SEO checklist provides a useful finding format. Restrict the initial review to important public pages; do not make private material public just because a visibility tool cannot fetch it.
Could security settings be blocking legitimate requests?
A security setting is worth checking when there is evidence that it affected a relevant request. Do not start by turning protection off. Ask the developer to verify the request against current official details and examine the specific rule involved. Any approved adjustment should be narrow, recorded and checked afterward so the access change does not create an unrelated security problem.
Perplexity's documentation includes crawler details intended to help site owners assess access. Use current published details rather than copying an old list of addresses from a forum. Official crawler guidance
Is the important page content available to a visitor?
Inspect the actual page, not just its URL. Confirm that the public explanation loads, that it is understandable without a private account and that links lead to useful destinations. Record any difference between what you intended to publish and what the visitor receives. This is a content-access review, not proof that the page will be chosen as an answer source.
For a hypothetical consultancy, a service page containing only a title and a booking widget leaves little explanation to evaluate. I would improve the visible scope and suitability details for customers before commissioning a platform-specific duplicate of the page.
How should you record a citation observation?
Preserve the prompt, date, product context, response and exact destination URL. Note whether the answer links to your page, names the business without linking or contains neither. Keep those categories separate. If the destination changes or the answer addresses a different task, record the difference instead of combining all appearances into a single unexplained number for the business owner.
| Review item | Record to retain |
|---|---|
| Access finding | URL, request context and response |
| Answer observation | Prompt and full response context |
| Linked source | Exact page, not just domain name |
| Follow-up action | Owner, reason and verification step |
The AI visibility measurement guide explains a reusable approach. The table above is a suggested record structure, not an actual set of results.
What should you conclude when access works but no citation appears?
Conclude only that the checks you performed did not identify the tested access problem, and that the observed answer did not cite the page. Those statements do not establish why the source was absent. Review task relevance and useful page details before making a further hypothesis. Avoid prescribing a large rewrite solely because one prompt produced a different set of sources.
If a provider offers a guaranteed citation after “unblocking AI,” ask how the claim follows from the evidence. For a practical review, discuss AI Search Optimization with a specific page and question in mind.
Frequently asked questions
Is a user-initiated page visit the same as an automatic crawl?
No. Perplexity documents user-initiated fetching separately from its crawler. Preserve that distinction when interpreting logs and access behavior. A request made because someone asked the product to visit a page is not, by itself, proof of how automated discovery will treat the site. Match the test to the question instead of treating every visit as the same evidence. Perplexity documentation
Should I disable my firewall to allow an AI crawler?
I would not recommend disabling the entire firewall as a first step. Establish whether a legitimate request was blocked and identify the narrow rule responsible. Have the person managing security evaluate any adjustment, document it and verify the result. Improving access to public content should not require exposing private resources or removing protections unrelated to the actual problem.
Does allowing a crawler guarantee inclusion in an answer?
Do not interpret access as a guarantee of selection. Record a successful access check as that specific result, then assess any actual citation separately. A provider should be able to explain the difference. If inclusion is promised, ask what evidence supports the promise and whether the work would still be useful to customers if the expected answer never appears.
Can two prompts produce different sources?
Compare the actual results rather than assuming every wording represents the same task. A prompt asking for a definition may need different details from one asking for a local provider. Keep prompts and response context with the record. If you want a repeated observation, use a consistent method and report any differences instead of selecting only a preferred answer.
Should I make a duplicate Perplexity-only version of my website?
I would first identify a reader need that the existing pages do not serve. A duplicate created solely for a platform label can add maintenance work without improving the explanation. Prefer a clear, accessible public page and a specific review of any access issue. Create new content when the task requires a distinct answer, not merely a different destination name.
Put it into practice
Illustrative example · not a client result
A security rule blocks a public service page. A developer can investigate that rule without switching off the entire firewall. Fixing access still does not promise a citation.
Your next step
Ask for the affected URL, request time and response before approving any security change.
Key takeaways
- Check Perplexity access and actual citations as separate tasks.
- Ask for the affected URL, request time and response before approving any security change.