Perplexity SEO Audit: Step by Step

A Perplexity SEO audit checks whether PerplexityBot can reach your site, which sources Perplexity cites for your buyers' questions, and why your pages are or aren't among them. Because Perplexity shows numbered sources on almost every answer, it is the most transparent AI engine to audit. Expect two to three hours for a first pass on 30 prompts.

What you need: a Perplexity account (free works), a spreadsheet, access to your robots.txt and, ideally, server logs.

Step 1: Can PerplexityBot crawl your site?

Perplexity documents two user agents (Perplexity):

Check three layers:

1. robots.txt. Look for User-agent: PerplexityBot groups and for broad User-agent: * disallows that cover key paths. A safe pattern:

txt User-agent: PerplexityBot Allow: /

1. CDN/WAF. Bot-management tools may challenge or block unfamiliar agents. Check your firewall events for PerplexityBot. If you need to allowlist by IP, use the ranges Perplexity lists in its crawler documentation rather than a third-party list.
2. Server logs. Filter for the PerplexityBot user agent. You want 200 responses on your key pages, not 403s or challenge pages.

Pass: PerplexityBot receives 200s on your top 20 URLs.

Step 2: How do you build the prompt set?

Use 30 prompts split across intent types. Pull wording from sales calls, support tickets and community threads, not just keyword tools.

Save the list with a version date so future audits use the same prompts.

Step 3: How do you run and log the prompts?

For each prompt:

1. Open a new thread (don't chain prompts; follow-ups change context).
2. Use the same mode and model choice every time and note it.
3. Run it twice, on different days if possible.
4. Log: brands mentioned (in order), whether you're recommended, every source URL, and which source numbers sit next to sentences about you.

Example row for format only.

Step 4: Which domains does Perplexity trust in your category?

Pivot the "all cited domains" column. Count how often each domain appears across all runs, then sort.

You'll usually see four groups:

• Review and comparison sites
• Community sources (Reddit, niche forums, Stack Exchange)
• Competitor sites (often their comparison or "alternatives" pages)
• Publishers and blogs in your niche

For the top 15 domains, record whether each mentions you and how.

Step 5: Why are competitor pages cited instead of yours?

Pick the five prompts that matter most commercially. For each, open the top cited page and your closest page side by side and score them:

The gaps are your rewrite brief. A common finding: the cited page is a plain, specific "X vs Y" or "best X for Y" page, while yours is a broad features page.

Step 6: What goes on the fix list?

Group fixes by type and owner:

1. Access (dev): robots.txt, firewall rules, rendering.
2. Pages (content): rewrite or create pages for the five priority prompts, answer-first, with tables and dated facts.
3. Sources (PR/community): for each top domain that omits you, define the action, such as a review drive, a correction email to an editor, or genuine participation in a thread.
4. Measurement (analytics): add perplexity.ai to your GA4 AI channel (tutorial).

Re-run the same prompts 30 days after the main fixes ship. Report citation rate, mention rate and recommendation rate against the baseline.

How do you present the audit to stakeholders?

Keep it to four slides:

1. Access status: pass/fail for PerplexityBot on key URLs, and any firewall fixes made.
2. Baseline: mention, recommendation and citation rates for you and two competitors across the 30 prompts.
3. Source map: the top 15 cited domains, with a tick or cross for your presence. This is usually the slide that gets budget, because it turns "AI visibility" into a concrete list of sites.
4. Fix plan: the top 10 actions with owner, effort and expected effect, labelled as expectations rather than forecasts.

Attach the raw sheet so anyone can check a cell against the original answer.

What does a good result look like?

There is no universal benchmark, and we won't invent one. Judge success against your own baseline and two named competitors on the same prompts. A meaningful audit outcome is a list of the domains that drive answers in your category, a ranked fix list, and a baseline you can re-measure.

Audit checklist

• [ ] PerplexityBot and Perplexity-User not blocked in robots.txt
• [ ] No firewall challenges for PerplexityBot
• [ ] 30-prompt set saved with version date
• [ ] Each prompt run twice; all sources logged
• [ ] Top 15 cited domains mapped with presence flag
• [ ] 5 priority prompts with page gap scores
• [ ] Fix list with owners and dates
• [ ] GA4 AI channel includes perplexity.ai
• [ ] Re-run scheduled 30 days after fixes ship, using the same prompts, modes and run count

Sources

• Perplexity, Perplexity crawlers
• Google Search Central, Introduction to robots.txt (general robots.txt syntax)
• IETF, RFC 9309: Robots Exclusion Protocol
• Bing, Which crawlers does Bing use?