Search intent: how it decides the format before you write

Search intent is the decision you make before writing, not a label you attach at the end. It sets the format of the text, and the wrong format reaches the wrong person even when the topic is right.

Intent is a decision, not a label

Many teams treat intent as a tag stuck on a keyword after the research is done. By then it changes nothing, because the draft is already written.

Used as a decision, intent answers three questions before the first line. Who is asking. What they already know. What they will do with the answer.

Those answers pick the format. A definition, a comparison, a step-by-step guide, a list and a product page aren’t interchangeable containers for the same text.

Google describes search in three stages: crawling, indexing and serving (Google, How Search Works). Intent lives in the third stage, where a query is matched to a page.

The same documentation doesn’t define intent as a separate concept. Better to say so than to quote a rule that doesn’t exist.

The intent of the query and the intent of the person

A query is a short text typed in a box. The person behind it has a goal the query may not show.

The same question serves different goals. A question about the cost of a static site can come from someone comparing proposals and from someone checking a budget.

The query looks identical in both cases. The format that helps doesn’t.

That is how a page can rank for a term and still fail. It got the words right and the goal wrong.

What each classic intent expects

Intent	What the person wants	Format that answers it
Informational	To understand something	Definition, explanation, example
Commercial	To choose between options	Comparison, criteria, table
Transactional	To complete an action	Product page, price, next step

The table is a starting point and not a rule. Real queries often sit between two rows.

Where the label fails

Three failures are common, and none of them is fixed by a better keyword list.

An ambiguous query. The same word can name a planet, an element and a brand. The label depends on context the query doesn’t carry.

A query that changes with the moment. The same person searches for a symptom before a diagnosis and for a treatment after it. The intent moved, and the page didn’t.

A query that mixes intents. Someone reading about a technique usually wants to apply it, so the informational page has to end in a usable step.

Nielsen Norman Group found the same instability from the behavior side. People turn to AI while the goal is still vague and to search when accuracy matters (NN/g, GenAI for Complex Questions, Search for Critical Facts). The tool follows the state of the question, and that state changes. The label on the link itself is a separate problem, and navigation and link labels covers what the research measures there.

How to check that you read the intent right

Two quick, free checks catch most of the errors.

Read what already ranks. The results on the first page show the format the engine currently treats as a match. If every result is a comparison and your draft is a definition, the intent was read wrong.

Watch what the person does after arriving. If people leave a definition page fast when it promised a price, the page didn’t give them what they came for. A scroll to the end of a guide that answered in the first paragraph is a different sign.

Google’s guidance asks whether the content provides a substantial and complete description of the topic (Google, Creating helpful content). That question only has an answer once the intent is clear.

What changes with AI search

An AI answer starts from a question written in natural language, which removes part of the guessing. It doesn’t remove the intent problem, and the research points in two directions at once.

Where the sources agree. Natural language lowers the keyword barrier, especially when the need is still vague (NN/g, How AI Succeeds and Fails to Help People Find Information). Traditional search stays for facts that need accuracy and trust (NN/g, GenAI for Complex Questions, Search for Critical Facts). Habits are slow to change, and people still start where they always started (NN/g, How AI Is Changing Search Behaviors).

Where the sources diverge. One reading splits the work by task type, with AI for complexity and search for critical facts. The study on information foraging, meaning how people hunt for information, splits it by what the person knows, since many don’t know how to prompt for what they need. The behavior is the same, but the cause isn’t.

A second divergence is about usability. The study of Google AI Mode reports discoverability and navigation problems, especially for people new to the tool (NN/g, Google AI Mode: Powerful Search, Poor Usability). The study of changing habits reports the opposite pressure, where familiarity keeps the old engine in place. Capability and usability aren’t the same measurement.

Google states that there is no additional requirement to appear in AI Overviews or AI Mode and no special optimization to write for them (Google, AI features and your website). The same page describes query fan-out, where one question becomes several searches.

That detail matters for intent. A fan-out expands the question into subtopics, so a page that answers one subtopic clearly is easier to use than a page that covers everything at once.

The fundamentals that still apply are collected in the search starter guide (Google, SEO starter guide).

Why this is a content decision

Intent decides the shape of the text, and the shape is chosen before the keyword is placed.

In practice, the test is quick. Write the format that answers the goal, then check whether the query would find it.

If the format is wrong, no amount of keyword placement repairs it. The reader arrives, reads two lines and leaves.

The related work on structure and extraction is in how AI answer engines choose and cite your sources. The cost of fixing the format late appears in design decisions before the build.