Synthetic query
A synthetic query is a web search an AI platform writes and runs by itself while answering someone. No person typed it.
A synthetic query is a search that an AI platform writes and runs by itself. Someone asks one question. To answer it, the platform makes up several searches of its own, runs them, and reads the results. Nobody typed those searches. The platform did.
That matters because your pages are now competing in searches no person ever made. Keyword tools only show what people type, so they miss these entirely.
An example
For example, someone asks: "What's a good coffee subscription for an office of 20 people?"
The platform might search for "office coffee subscription UK," "coffee subscription 20 people price," "bean to cup office coffee" and "Fernhill office coffee review." Then it reads the top results for each and writes one answer.
If Fernhill Coffee only has a page that ranks for "coffee subscription," it may miss most of those searches. If it has a clear page on office coffee with prices by team size, it has a chance in several of them.
Why they're hard to plan for
Synthetic queries change from one day to the next, and from platform to platform. The same question can lead to different searches on Monday and Tuesday, so the pages read and the brands named change too. That's one reason answers vary.
They also tend to be more specific than what people type. Platforms add details like a country, a price or a brand name, because they're looking for pages that answer the question exactly.
In AnswerOrbit
When a platform shows the searches it ran, we keep them. Open an answer in Answers to see the web searches behind it. The Searches tab in Sources shows how often each platform searched before answering, and how often you were named when it did. We keep searches from ChatGPT, Perplexity and Copilot.
For product questions, the Background searches tab in Shopping shows the shopping searches ChatGPT ran.