Search intent
What a person is actually trying to accomplish with a query, in Google's own four categories, and why an answer engine resolves it before it ever reaches your page.
Karl-Gustav Kallasmaa, Founder & CEOLast updated Search intent is what a person is trying to accomplish with a query, as distinct from the words they typed. Google states the premise directly in How Search Works: "To return relevant results, we first need to establish what you're looking for – the intent behind your query."
The SEO industry usually splits intent four ways — informational, navigational, transactional, commercial investigation. That taxonomy is useful shorthand, but it is not Google's. Google's own vocabulary is published in the Search Quality Rater General Guidelines, version dated 11 September 2025, and it is worth using the real terms because they are more precise.
Google's four categories
The guidelines tell raters to think of queries as having one or more of the following intents:
- Know — the intent is to find information or explore a topic. Within it, Know Simple queries seek a very specific answer, such as a fact or diagram, which "has to be correct and complete, and can be displayed in a relatively small amount of space." The stated rule: if most people would agree on a correct answer, and it would fit in one to two sentences or a short list, it is Know Simple.
- Do — the intent is "to accomplish a goal or engage in an activity," which may mean to download, buy, obtain, be entertained by, or interact with a site or app.
- Website — the user is looking for a specific website or webpage.
- Visit-in-person — the user wants a business, an organisation, or a category of businesses they can go to.
Two details in the guidelines matter more than the list itself. Google explicitly says the line between Do and Know is "sometimes blurry" and tells raters not to worry about strictly differentiating them. And most queries are not Know Simple: the guidelines name broad or in-depth informational queries, ambiguous ones, controversial topics, queries with no definitive right answer, and queries where users want to browse, find inspiration, or read personal opinions.
Interpretations are graded separately from intents
The guidelines keep a second, orthogonal vocabulary for ambiguity. A dominant interpretation is "what most users mean when they type the query" — and Google notes that not all queries have one. A common interpretation is what many or some users mean, and a query can have several. A minor interpretation is one fewer users have in mind, split into reasonable minor interpretations, which help fewer users but are still legitimate, and unlikely minor interpretations, which are possible in theory but not worth serving.
Google's worked example is the query harvard: depending on need and location, users may want the official homepage (Website), directions (Visit-in-person), or information about the school (Know) — while a specific individual Harvard course is called out as an unlikely interpretation of so broad a query.
Why it decides whether you are cited
Because an answer engine resolves intent before your page is consulted, not after.
In a classic ranked list, a page matched to the query string but not to the goal still appears, and the user decides. In a conversational or AI-generated answer, the system reformulates the request into retrieval queries against the inferred goal, fetches passages, and drops anything that does not answer it. A page that is topically adjacent is retrieved and then discarded silently. There is no position four to fall back to.
Know Simple queries are also the ones most likely to end without a click at all, since the answer fits in the surface. That is the ordinary case of zero-click search and of featured snippets before it, and it is not a defect to be routed around — it is the intent being satisfied.
The failure modes
- Answering a Know query with a Do page. The guidelines include this exact case: for a query seeking health information about egg salad, recipes with no nutritional information are "not a reasonable minor intent," because "the user issuing the query is specifically looking for health information, not food preparation."
- Ignoring important words. Google's other worked example is a query for a non-emergency number, where returning the emergency number is judged unreasonable, with the note that "users can become frustrated when results ignore important words."
- Serving one page to several intents. A query with two common interpretations needs two pages. A single page that hedges across both matches neither well enough to be the retrieved passage.
- Assuming a dominant interpretation exists. Google says some queries have none. Building a page around a guess at "what people really mean" produces content for an intent a minority holds.
- Treating intent as static. The guidelines note that as new products appear, the dominant interpretation of a query changes again.
How to act on it
- Classify before writing. Know, Know Simple, Do, Website or Visit-in-person — and write the page the category actually calls for.
- For Know Simple, put the answer in the first sentence, correct and complete, and short enough to be lifted whole.
- Split, don't hedge. Two common interpretations are two pages, each of which states its interpretation in its opening line.
- Honour the qualifying words. "Non-emergency", "without", "free", "for beginners" are the intent. A page that quietly drops them answers a different query.
- Re-check periodically. Intent for a query drifts as the underlying world changes; long-tail queries drift fastest.
Terms related to Search intent
Longer, more specific queries that each occur rarely but together make up most of what people ask — why the tail exists mechanically, and why it is the part of demand an assistant answers directly.
Google's reversed search result, where a passage lifted from a page is shown above the link — selected by Google's systems, requestable by nobody, and refusable with one meta rule.
A search that ends without the reader visiting any website, and the measured gap between sessions that show an AI summary and those that do not.
Search where the query is a turn in a conversation rather than a standalone string — how the earlier turns get carried, and what that does to the passage of yours that gets retrieved.
Google's AI-generated summary at the top of a results page, and the snippet controls that decide whether your page can appear inside one.