Search intent is the goal behind a query: what the person is trying to get done, as distinct from the words they typed. Google's rater guidelines define user intent as what someone is trying to accomplish when they search, and note that a single query can carry several intents.
How search intent works
The classic categories come from Andrei Broder's paper "A taxonomy of web search", published in SIGIR Forum in fall 2002 and based on his work at AltaVista.
| Intent | The searcher wants to | Share of queries |
|---|---|---|
| Navigational | reach a particular site | 20% |
| Informational | find information on one or more pages | 48% |
| Transactional | do something through the web, such as shop or download a file | 30% |
Queries that were neither navigational nor transactional were counted as informational, and Broder called the log figures "soft", because reading intent from a query is inexact.
Google's Search Quality Rater Guidelines use a finer set: Know queries (including Know Simple queries with short factual answers), Do queries, Website queries and Visit-in-person queries. They also add two ideas. A query can have several interpretations, from dominant down to minor, and the dominant one can change over time: [iphone] meant the first model when it launched in 2007 and now usually means the most recent or upcoming one.
Why search intent matters for AI answers
The kind of query shapes whether an AI answer appears at all. Google says AI Overviews show only when its systems judge them additive to classic search, so they often don't trigger. In Pew Research Center's March 2025 data, 60% of Google searches that began with a question word such as who, what, when or why produced an AI summary, against 8% of one- or two-word searches.
AI features also break a query into parts. Google says AI Overviews and AI Mode may use query fan-out, running related searches across subtopics. Its example: "how to fix a lawn that's full of weeds" might fan out to "best herbicides for lawns", "remove weeds without chemicals" and "how to prevent weeds in lawn". Google adds that its AI systems can judge a page relevant even when it does not match the query's exact words. Microsoft says AI systems grounded in search indexes weigh how clearly each page satisfies the intent behind a query.
For AI visibility, the practical step is to track prompts that reflect real intents, not just keywords; see how to choose the prompts to track.
An example: [mercury]
Google's rater guidelines use the query [mercury]. The planet and the chemical element are common interpretations, an insurance company is a reasonable minor one, and a British aircraft designed during the Second World War is an unlikely one. A results page, or an AI answer, that serves only one reading fails the people who meant another.
Common misconceptions
- "Every query has one intent." The rater guidelines say one user can have several intents for a single query. Their example [walmart] could mean finding a nearby store, shopping online or reading about the company.
- "An intent label is a fact." It is an inference. Broder, who proposed the categories, stressed that intent cannot be read from a query with certainty.
- "Cover every intent with a page per phrasing." Google says making separate content for every variation of how people search, including fan-out queries, primarily to manipulate rankings or AI responses violates its scaled content abuse policy. Its AI systems understand synonyms.