Perplexity
citation audit.

Perplexity runs a live web search for every query, fetches the top results, and synthesises an answer with paragraph-level citations. The sources it tends to favour:

• High organic-search authority on the query. Position-1 to position-10 in Google for the query, or close synonyms.
• Specificity of the answer. Pages that directly address the query in their first paragraph get cited far more often than pages that introduce the topic before answering.
• Recency. Perplexity’s relevance score visibly weights freshness. A 2024 article on an AI topic is now treated as historical context. A 2026 article is treated as primary.
• Source diversity. Perplexity tries to cite three to six distinct domains per answer. It dislikes citing the same domain twice unless it has to. That means there’s a citation slot for niche specialists, not just the dominant publishers.
• Structured signals. Schema, especially Article and FAQPage, makes content cheaper for the answer layer to extract. We covered the structural mechanics in The Citation Graph Doesn’t Read Your Drafts.

Perplexity is the easiest of the three to audit precisely because the citations are first-class UI. Every other engine requires some inference. That is why the Perplexity audit is the cheapest visibility diagnostic available in 2026, and why it is also the most punishing: there is no plausible excuse for not knowing what your sector’s answers look like in it.

This is the method we use for every F&T client. Forty-five minutes, no tools beyond a browser and a spreadsheet.

1. List the queries that matter to your business. Ten to fifteen. Mix branded (your name + category), commercial (category + location), informational (“how to”, “what is”), and comparison (“X vs Y”).
2. Run each query in Perplexity. Read the answer. Open the Sources panel. Note: which domains are cited, which paragraphs cite which sources, whether you’re cited at all.
3. For each non-citing query, ask one question: what does the cited source have that we don’t? Open it. Read it. Note the answer format, schema, freshness signals, author credentials, fact density.
4. For each citing query, capture the citation context. Which paragraph cites you? What claim does it support? Is the claim flattering, neutral, or off-brand? AI engines occasionally cite you for the wrong reason.
5. Tally the results. Out of fifteen queries, how many cite you? On which themes are you over-represented or under-represented? Where are competitors winning?

From dozens of audits, the diagnosis usually reduces to one of five patterns:

• You’re not on page 1 of organic search for the query. Perplexity’s shortlist is broadly the top organic ten. If you’re on page 2 or below, this is a Google ranking problem first.
• Your page rambles before answering. The answer layer scans the first 300 to 500 words. If your answer arrives at word 700, the layer extracts a competitor’s and cites them.
• Your page lacks freshness signals. No visible publish date, no “last updated” line, no schema datePublished, no recent inbound links. Perplexity treats stale-looking content as historical.
• You’re thin on the specific sub-topic. Your sector page covers ten themes shallowly. The query is about one of them. A specialist domain that has a dedicated 1,500-word page on just that theme gets cited instead.
• Robots.txt blocks PerplexityBot. A surprisingly common finding via Cloudflare WAF defaults. The fix is one line in robots.txt or one toggle in Cloudflare. Always check first.

Working from the diagnosis, the order matters:

1. Unblock PerplexityBot. Free, immediate. If blocked, every other fix is wasted effort.
2. Restructure first-paragraph answers on your top ten target pages. Lead with the answer to the query the page targets. Move the brand intro lower.
3. Add visible publish and updated dates. In the body, not just in schema. The answer layer reads what humans read.
4. Build dedicated specialist pages on the sub-topics where you’re currently thin. One 1,500-word page on a narrow theme often outperforms a 600-word page on the broad theme.
5. Earn placement in the third-party citation graph. Trade press, directories, comparison sites, Reddit. Perplexity cites third-party hosts as readily as first-party. Sometimes more readily.

Perplexity citations move week to week. The honest reporting cadence for a small or mid-size UK business is monthly: re-run the same fifteen queries, note the deltas, decide whether the pattern is moving the right way.

The wrong cadence is daily: noise dominates. Citation rates fluctuate with the freshness window, with new competitor publications, with Perplexity’s own model updates. Daily reporting will whip you around. Monthly is enough to see real movement.

For a quarterly or annual board update, supplement the citation count with the qualitative read: are we cited for the right claims? Are we cited alongside the right peers? Has any competitor become disproportionately cited on a query we used to own? The data answers the count. The reading answers the why.