AI Brand TrackingHow ChatGPT and Perplexity run query fan-outs.
When an AI answers a question, it quietly issues its own background search queries — the query fan-out — to gather sources. Vercite captured and analysed 50,307 query fan-outs straight from our own platform data. The two engines behave nothing alike: ChatGPT explores a topic with long, rephrased, often translated sub-queries; Perplexity looks up an answer with a few short keyword strings in the local language.
The searches you never see.
An AI answer is built on searches nobody typed. Understanding that hidden middle step is the whole game: your brand isn’t competing for the user’s question — it’s competing for the queries the engine derives from it.
A user types one prompt into the AI — in their own words, their own language.
Before answering, the engine silently issues its own background search queries. That set is the query fan-out.
Each fan-out query pulls sources from the web. Whatever ranks for those strings becomes the raw material.
The reply cites the brands and pages the fan-outs surfaced — not the ones that rank for the original prompt.
One explores, one looks up.
Side-by-side profile of how each engine generated its background queries across our full fan-out dataset.
Rephrases the whole question, frequently translates it to English, and probes named candidate brands. A broad, low-repetition set of search queries.
Issues a handful of short keyword queries in the prompt’s own language, then logs each once per source it pulls. Heavily templated by entity × city.
Raw row counts lie.
Both engines log a similar number of rows per prompt, but most of Perplexity’s are exact repeats. The chart shows distinct queries per prompt once duplicates are removed — the true measure of how widely each engine searches.
Perplexity repeats the same ~3 strings up to 165 times per prompt. Always dedupe it before reporting breadth.
ChatGPT works the funnel. Perplexity stays at the top.
Share of each engine’s distinct fan-outs that signal a given intent. ChatGPT spreads into definition, comparison and reputation questions; Perplexity concentrates on best-of and location. A query can carry more than one intent.
ChatGPT rewrites you into English.
Share of distinct fan-outs issued in a different language than the original prompt. ChatGPT translates four in ten background queries — mostly Nordic prompts pushed into English — while Perplexity keeps nine in ten in the prompt’s own language.
Translates local queries before searching — English content can win even for Swedish prompts.
Searches in the prompt’s own language — local-language pages are what get cited.
Reuse your words, or reinvent them.
How closely each background query matches the words in your original prompt. Near-verbatim means the fan-out barely changes the prompt; fully reformulated means almost every word is new. ChatGPT mostly rewrites; Perplexity mostly trims.
A rich vocabulary vs one big word.
The intent words that recur most across each engine’s distinct fan-outs — ranked by how widely they span brands, then folded across languages into one label. Word size = frequency. ChatGPT reaches for a wide spread; Perplexity leans almost entirely on “best”.
Long and curious vs short and repetitive.
Three shape metrics that sum up the difference: query length, how often a fan-out is phrased as a real question, and how concentrated the vocabulary is.
One brand needs two content strategies.
Try Vercite for free today
Monitor your brand across ChatGPT, Gemini, Perplexity & Google AI