Search Intent: Types, SERP Reading and AI Prompts

Search intent is the goal behind a query: what the person typing it wants to learn, do, reach or buy. Google ranks the pages that best satisfy that goal, so a page written for the wrong intent will struggle to rank for a keyword however good it is. Most SEO problems that look like “not enough content” or “not enough links” turn out to be intent problems.

This guide is for business owners and marketers who pick keywords and brief content. It covers the four standard intent types and the modifiers that change them, what Google’s own rater guidelines call intent, how to read a results page before you write, the symptoms of a mismatch in Search Console, and how intent shows up in the longer prompts people type into AI assistants.

On this page

• What is search intent?
• The four types of search intent, plus modifiers
• What Google’s rater guidelines call intent
• How do you read the SERP for intent?
• Symptoms of a search intent mismatch
• How to fix a page that misses the intent
• Search intent in conversational AI prompts
• Where search intent fits in your SEO plan

What is search intent?

Search intent is the reason a query was typed, and it decides what kind of page can rank for it. Two keywords can share almost every word and still want completely different pages. “Accounting software” is a shopping query: the results are vendor pages and comparison lists. “Accounting software journal entries” is a how-to query: the results are help articles and tutorials.

The practical consequence is simple. Before you ask “can we rank for this keyword,” ask “what page would satisfy the person searching it, and are we the business that should publish it.” If the honest answer is a product page and you were planning a blog post, stop there. Keyword volume tells you how many people search. Intent tells you whether any of them want what you are about to write.

Intent is also why keyword research for SEO cannot be done from a spreadsheet alone. Tools estimate intent from patterns in the words. The results page shows what Google has actually decided.

The four types of search intent, plus modifiers

The four types used across the SEO industry are informational, navigational, commercial and transactional. Semrush’s search intent guide defines them the standard way:

• Informational: the searcher wants to learn about something. “What is a 1031 exchange,” “how to descale a kettle.”
• Navigational: the searcher wants a specific site or page. “Gusto login,” “IRS payment portal.”
• Commercial: the searcher is researching options before a purchase. “Best payroll software for restaurants,” “Gusto vs ADP.”
• Transactional: the searcher wants to act now. “Buy standing desk,” “book dentist Saturday.”

The four boxes are a starting point, not a law. Real queries carry modifiers that shift or stack intent, and those modifiers matter more than the category label. The diagram below shows the four types with the words that usually signal them and the page type that usually wins.

Scroll the diagram sideways to see all of it.

The modifiers worth watching:

• Location words (“near me,” a city name) add a local layer. The map pack appears, and a national guide rarely competes with it.
• Years and “latest” add freshness. Results skew to recently updated pages.
• Audience words (“for dentists,” “for beginners”) narrow who the answer is for. A generic page loses to a specific one.
• Comparison words (“vs,” “alternative to”) turn an informational topic commercial.
• Price words (“cost,” “cheap,” “pricing”) pull the query toward buying, even when the phrasing sounds like a question.

What Google’s rater guidelines call intent

Google’s own vocabulary is different from the industry’s four boxes, and it is more useful for thinking about pages. The Search Quality Rater Guidelines (the September 11, 2025 edition) ask raters to think of queries as having one or more of these intents:

• Know: the user wants to find information or explore a topic. A subset, Know Simple, wants one correct answer that fits in a sentence or two or a short list, such as a height or a date.
• Do: the user wants to accomplish a goal or engage in an activity: download, buy, obtain, be entertained, or interact with a site or app.
• Website: the user wants a specific website or page.
• Visit-in-person: the user wants a physical business or a category of businesses nearby.

Two parts of the guidelines are worth borrowing directly. First, the guidelines say the line between Know and Do is “sometimes blurry” and tell raters not to worry about strict classification. Neither should you. Second, raters judge a query’s dominant interpretation (what most users mean), common interpretations (what many or some users mean) and minor interpretations. The guidelines use [iphone] as an example of a query whose dominant meaning changes over time as new models ship.

That second idea explains a common frustration. A page can match a real intent and still rank poorly because it serves a minor interpretation. If most people searching a phrase want the software and you wrote about the concept, you are competing for the leftovers.

How do you read the SERP for intent?

You read intent by looking at what already ranks, because the results page is Google’s current best guess at what searchers want. Keyword tools label intent from word patterns. The SERP shows the outcome of Google testing pages against real behavior.

Ahrefs’ “3 Cs” is a tidy way to structure the reading. In Ahrefs’ search intent guide, content type is the dominant kind of page (blog post, product page, category page, landing page, video), content format is the dominant shape (how-to, list, review, comparison) and content angle is the hook the top results share. The flow below is how we apply it.

Scroll the diagram sideways to see all of it.

A few reading habits that make this more accurate:

1. Search logged out, in the right country. Personalization and location change results, especially for anything with local intent.
2. Ignore the one outlier. A single strong domain ranking with an off-type page is usually riding its authority, not revealing the intent.
3. Treat SERP features as intent evidence. A map pack says Visit-in-person. A video carousel says people want to watch. A People Also Ask box lists the follow-up questions your page should answer, and a featured snippet tells you the format Google likes for the core answer.
4. Read the AI Overview. It shows which sub-questions Google thinks belong in a complete answer.

Symptoms of a search intent mismatch

An intent mismatch shows up in your data before anyone diagnoses it. The page gets impressions because it is relevant to the words, but it never climbs, or it climbs and nobody clicks. Search Console is where to look: the Performance report gives clicks, impressions, CTR (clicks divided by impressions) and average position per query and per page, and our Google Search Console guide walks through the filters.

The CTR symptom has its own mechanics, covered in detail in CTR in SEO.

How to fix a page that misses the intent

Fixing a mismatch means changing the page to match what the searcher wants, or changing the keyword the page targets. Polishing the prose is rarely enough. In order of effort:

1. Retitle. If the content is right but the title frames it wrongly (“Our CRM features” for a comparison query), rewrite the title and meta description to the dominant intent. This is the cheapest test.
2. Restructure. If the format is wrong, change it. A comparison query wants a table early. A how-to wants numbered steps. A Know Simple query wants the answer in the first sentence.
3. Retarget. If the page is genuinely good at an adjacent intent, point it at that keyword and stop fighting for the original one.
4. Rebuild or split. If the query wants a page type you do not have, such as a category page instead of an article, build that page. If one page tries to serve both learners and buyers, split it into two and link them.

What does not work is adding 1,000 words to a page of the wrong type. Google’s own guidance says there is no preferred word count, and length does not convert a guide into the product page a transactional query wants.

Search intent in conversational AI prompts

In AI assistants, intent is usually stated rather than implied, which makes it easier to read and harder to fake. A searcher types three words into Google. The same person types a paragraph into ChatGPT, Gemini or Google AI Mode, with their situation, constraints and task spelled out. The diagram below breaks one prompt into parts.

Scroll the diagram sideways to see all of it.

Three things change when intent arrives as a prompt:

• The engine searches more than once. Google’s AI features documentation says AI Overviews and AI Mode may use a “query fan-out” technique, issuing multiple related searches across subtopics and data sources. Each constraint in the prompt can become its own sub-search, and your page competes for those sub-questions, not only the head term.
• The mix of intents is different. A September 2025 NBER working paper on how people use ChatGPT found that practical guidance, seeking information and writing together make up nearly 80% of conversations, and that non-work use grew to over 70% of messages. People bring advice-seeking and decision-making to assistants, not just lookups.
• Situational pages win passages. A page titled “CRM for trades businesses that already use QuickBooks” answers the constraint directly. A generic “best CRMs” list makes the model do the work of filtering, and it may pick a source that already did.

If you want those answers to name you, see how to rank in ChatGPT search and Google AI Mode explained. The structural side, writing sections that survive being quoted alone, is covered in content chunking for AI.

Where search intent fits in your SEO plan

Search intent should be checked before a page is written, not diagnosed after it fails. In practice that means one extra column in your keyword list (the page type that ranks today), one habit before every brief (read the top ten and the AI Overview), and one monthly check in Search Console for the mismatch symptoms in the table above.

It also changes how you plan a site. Each intent wants its own page type, so a healthy site has explainers for Know queries, comparison pages for commercial ones, and product, service or booking pages for Do queries, linked so a reader can move from one to the next. That structure is the base for topical authority, and it is the same structure AI assistants use when they pull one passage for context and another for the recommendation.

If you want a second pair of eyes on which of your pages are fighting the wrong intent, and how your content maps to the prompts customers now type into AI assistants, that is part of our AI SEO service.