How Can I Check Whether Perplexity Found My Page but Excluded It From the Answer?

Why does the difference matter?

"Not cited" has two very different causes:

• Not found: Perplexity's search did not retrieve your page for the question. The fix is about crawl access, indexing and ranking for that query.
• Found but excluded: your page was retrieved, but the answer drew on other sources. The fix is about the page itself: whether it answers the question clearly, whether it can be read, and whether other sources look more useful or more trustworthy.

Working on the wrong one wastes effort. Adding more content to a page Perplexity never retrieves changes nothing. Chasing links for a page that is retrieved but unreadable to Perplexity's fetcher also changes nothing.

Check 1: What does the answer view show?

Run your target question in Perplexity several times, since sources vary between runs. For each run, record:

• The sources listed for the answer, and which of them are cited next to specific claims.
• The searches shown in its more thorough modes, which reveal how Perplexity rephrased your question.
• Whether your domain appears anywhere, even for a different page than the one you expected.

If your page appears in the sources but no claim is attributed to it, it was found and used lightly or not at all. If the searches Perplexity ran do not match the wording on your page, that is useful too. It tells you which phrasing you need to answer.

Check 2: What does the API show?

Perplexity's API gives more structured evidence, although its results can differ from the consumer app:

• Agent API responses (Perplexity's API since Sonar Chat Completions ended on 27 September 2026) include a search_results item listing each source retrieved, with title, URL, date, last-updated date and snippet, and the answer text marks the claims it cites with numbered references such as [1]. A URL in search_results that no numbered reference points to was retrieved but not cited.
• The Search API returns ranked results for a query, up to 20, with ten by default. Run your question and see whether your page appears, and where.
• A domain filter limits Search API results to domains you list. Run the question restricted to your own domain. If your page comes back with a relevant snippet, Perplexity can find and read it. If it comes back with a snippet of navigation text or an unrelated passage, the content Perplexity extracts is the problem.

Also look at the date Perplexity associates with your page in the results. If it shows an old date while competitors show recent ones, freshness may be counting against you.

Check 3: What do your server logs show?

Perplexity uses two crawlers. PerplexityBot builds its index and respects robots.txt. Perplexity-User fetches pages when a user's question needs them, and generally does not follow robots.txt rules, since it acts for a person.

• A Perplexity-User request at the time of your test means your page was fetched live for that question. If it was then not cited, it was excluded.
• A failed request, such as a 403 from a firewall or bot-protection service, a timeout or a redirect loop, means Perplexity tried and could not read it.
• No recent PerplexityBot visits suggest your page may not be in Perplexity's index at all.

Whether Perplexity needs PerplexityBot access to cite you is covered in whether Perplexity requires PerplexityBot access to cite a website.

How do you read the results?

Why would Perplexity exclude a page it found?

• The answer is buried. The passage Perplexity extracts does not contain the answer, because it sits deep in the page, in a tab, or after long introductions.
• The content loads with scripts. Most AI crawlers do not run JavaScript, so text rendered in the browser may be invisible to them.
• Other sources say the same thing with more authority, such as a well-known publication, an official body or a major review site.
• The page looks one-sided for a neutral question, as with a vendor's comparison page. How that is judged is covered in how to check whether a competitor-controlled comparison page is treated as neutral.
• The page looks stale, with an old date or outdated details, while competitors show recent updates.
• The page contradicts other sources without evidence that it is right.
• The page is behind a paywall, login or cookie wall that blocks the text.

How do you improve the odds of being used?

• State the answer early, in a sentence that could stand alone as a quote.
• Use the wording Perplexity searched with, as shown in its search steps.
• Add evidence others lack: original data, specific numbers, named methods or first-hand detail.
• Keep the page current, and show real update dates when you change it.
• Make sure the content is in the HTML, not injected by scripts.
• Allow Perplexity's fetcher through your security settings.

Then rerun the same checks. Whether a citation that does not support its claim should count is covered in whether to count a Perplexity citation when the linked page does not support the claim.

What might an illustrative case look like?

Consider an illustrative payroll software company whose guide to contractor tax forms never appeared in Perplexity's answers. Server logs showed Perplexity-User fetching the guide within seconds of each test, so it was being found. A Search API query restricted to the company's domain returned the guide, but its snippet was a cookie notice and navigation text. The answer sat in an accordion loaded by script. The company moved the answer into the page's HTML, opened with a two-sentence summary, and added a table of deadlines. Two weeks later, the guide was cited in most runs for the target question.

What do people ask about Perplexity excluding pages?

Does Perplexity show every source it read?

Not always. The interface shows the sources behind an answer, and the API shows retrieved results, but neither is a complete log of everything considered. Use them together with your server logs.

Is the API a reliable stand-in for the app?

It is useful for diagnosis, but answers and sources can differ from the consumer app. Confirm important findings in the app.

How many runs do I need?

At least three to five per question, since sources change between runs. One run can make a page look excluded when it was simply unlucky.