Engines

Perplexity

Perplexity AI · Perplexity Search

Definition

Perplexity is an answer engine that retrieves web pages and writes a cited response in the same window. For digital PR it is the loud citer: users see numbered sources, and practitioner crawls routinely find more outbound links per answer than ChatGPT. HubSpot, citing Fan Out, reported about 10.8 sources per Perplexity answer and a large share of off-site citations coming from this engine in that dataset. Your own logs will differ by category. The working habit is the same. Read the source list.

Because the product is built to show pages, an earned URL that is indexable and specific has a shorter path into the footnote than it does in a chat product that often answers without links.

How it works

A user asks a question. Perplexity pulls candidate pages, writes a synthesis, and attaches sources. Pages that state a number, a date, and a name are easier to attach than pages that sell. News stories, documentation, .gov primers, and methods pages show up constantly. Homepage slogans do not.

Example: a query “what is the median close time for Series A in 2025” returns an answer with eight sources: two venture-press features, one NVCA PDF, two law-firm memos, a Substack, and two owned “state of fundraising” landing pages. A seed-stage bank that commissioned one of the venture-press features is in the list twice — once as the news URL, once as its methods page. A rival bank with a prettier report and no press links is absent. The earned story did the retrieval work.

Track Perplexity on a fixed query set, logged-out if you can, and store the source URLs. Focus and Pro modes, follow-up turns, and personalization will drift the list. Note the mode in the sheet.

How it differs

ChatGPT may name you without a URL. Google AI Overviews mix generated text with supporting links inside Search and only fire on some queries. Claude is a separate retriever with its own refusal and citation habits. Calling all four “ChatGPT traffic” in a board deck erases the only engine that consistently hands you footnotes to audit.

Common errors

Optimizing a page for “Perplexity keywords.” Assuming a citation passes PageRank. Paying for a roundup that Perplexity then treats as a weak listicle. Forgetting that a competitor’s recap of your study can outrank your study in the footnote.

Sources