Does my website show up in Perplexity?

How do robots.txt rules affect Perplexity?

Perplexity publishes two agents worth separating:

• PerplexityBot builds the search index and follows robots.txt.
• Perplexity-User fetches when a person asks a question and generally ignores robots.txt because a user requested the fetch, per Perplexity's own documentation.

Blocking the first keeps you out of the index path you control with robots.txt. It does not shut every path by which a page can appear. The two agents are compared on PerplexityBot vs Perplexity-User, and the robots.txt generator writes the rules in the form operators expect.

Before you change rules, write down which agent you mean to allow or block. Teams often block a training crawler name they saw in a blog post while leaving the search crawler alone, or the reverse.

Which questions should you ask in Perplexity?

Search your own brand name and you will usually find yourself, because a branded answer can be assembled from a mention rather than from your page. Ask best [category] for [your buyer type] or how do I [job your page does] instead, and read the citations under the answer.

Build a fixed list of ten to twenty questions and keep it for a month. The useful number is how many of those questions cited your domain at least once, not whether you appeared in the first answer you tried on a Tuesday morning.

In our workspace, the meta-question "does my website show up in Perplexity" is itself tracked. In the 30 days to 4 October 2026, Perplexity returned 57 answers and attached 565 citations across them; none pointed at attensira.com. Agency and tool blogs dominated the deciding sources. That is a selection problem on a question we care about, not a crawl problem: PerplexityBot had already fetched the site.

If a page is never cited, which problem is it?

There are three, and the fix is different for each:

• Never fetched. The crawler has not read the page. Look at robots.txt, a WAF rule, or whether the text only appears after JavaScript runs.
• Fetched, not cited. The crawler read it and chose somebody else's page. That is a passage problem: the answer wanted one self-contained sentence it could lift and did not find one. Answer blocks are written for that failure mode.
• Cited, but not as you. Perplexity used a third-party page that mentions you. That is an off-site problem, not a page problem.

Attensira answers the first from the edge's own crawler readings, and the second and third from the answers themselves: which of your pages were cited for a tracked question, and which pages that should answer one were never fetched. The AI search visibility guide takes each of the three in turn.

How is Perplexity different from ChatGPT for this check?

Both are answer engines, not rank tables. ChatGPT's search features use OpenAI's documented search crawlers; Perplexity uses PerplexityBot for indexing and Perplexity-User for person-triggered fetches. The log signatures differ, and the source lists under answers differ in length and domain mix.

If you only optimize for one engine's crawl pattern, you can be reachable in Perplexity and invisible in ChatGPT search, or the reverse. Run the same buyer question set on both surfaces and read citations separately, with a sample size per engine. Analyzing log files shows how to split crawlers by purpose on your own site.

What does a healthy monitoring loop look like?

1. Weekly log pass. PerplexityBot requests and status codes on your priority URLs.
2. Daily or weekly answer read. The fixed buyer question set on Perplexity (and ChatGPT if that surface matters to you).
3. Join the two. When a page is fetched but never cited, rewrite the passage that should answer the question. When it is never fetched, fix access first.
4. Re-read after ship. Compare the same questions after the live page changes, with the answer count beside any rate movement.

Skipping step 1 makes step 2 noisy: you cannot tell access from selection. Skipping step 4 treats one lucky answer as a trend.

What are the common mistakes?

• One branded query. You learn whether Perplexity knows your name, not whether it uses your product page on category questions.
• Treating crawl volume as winning. Training crawlers can dominate request counts; PerplexityBot coverage on buyer URLs matters more for citations.
• Ignoring user-agent spoofing. Log a visit as "claimed PerplexityBot" until IP verification passes.
• Changing robots.txt without naming the agent. PerplexityBot and Perplexity-User follow different rules.
• Reading a single answer as a score. Perplexity's sources shift run to run; only a rate over many answers is comparable week to week.

How does Attensira check both reach and citation?

Attensira reads crawler traffic at the edge on your site and reads answers on a schedule for the questions you track. Crawler logs show which search crawlers fetched which pages; citation tracking shows which answers linked your domain. The crew uses the join: when a page that should answer a tracked question has never been fetched, that is the first fix; when it is fetched and never cited, the rewrite targets the passage.

Start with the public audit if you want a first read of what ChatGPT and Perplexity say about you today on a short question set, then promote the questions that matter into daily tracking.

What is the honest limitation?

Being crawlable does not mean being cited. Perplexity picks a small set of sources per answer and they vary from run to run, so a single check tells you almost nothing. Measure the share of relevant questions you appear in, over time, with the number of answers behind the rate, or you are reading noise as a score.

What should you write down before you change anything?

A one-page brief keeps teams from chasing the wrong lever:

• The buyer question verbatim, not a keyword-stuffed variant.
• The URL that should win, linked in your CMS.
• Last PerplexityBot fetch for that URL from logs, with status code.
• Last five Perplexity answers to the question, with domains cited.
• The hypothesis: access, passage or off-site mention.

Review the brief weekly. If fetch dates move but citations do not, you are in the passage bucket. If fetch dates are stale, stay in access until the crawler returns.

Save PDF or HTML exports of answers when you review them. Perplexity refreshes sources often; an export dated Monday is evidence when Friday's answer looks different. Evidence beats memory when you later judge whether a rewrite worked. Store exports beside the log lines from the same week so access and citation checks stay aligned.

When should you compare Perplexity to ChatGPT?

When both surfaces matter to your buyers. The crawlers differ, the source lists differ, and a page can win one engine while losing the other on the same wording. Run parallel question sets, store answers with dates, and report rates per engine with separate denominators. A combined "AI visibility %" across engines hides the tradeoffs you need to see when robots.txt, structure or PR changes are on the table.