Understanding Search Intent: Which Page Fits Which Search Query

Why do well-written pages often get no visitors?

A business has a service page rewritten. It is thorough, contains the important search term several times and loads quickly. Still it does not appear on the first results page. Instead, the top spots go to guides, comparison portals or map results with local businesses. The reason is rarely technical. Usually the kind of page does not fit what searchers want to achieve with this query.

This article answers the question: How do you recognize the intent behind a search query, and which page fits it? We first clarify the term and the best-known model in search research. Then we look at what studies say about the distribution of intents and about new search habits. At the end there is a simple test, a mapping table and a checklist for your own website.

The search query names the topic. The results page reveals the intent.

What is search intent? Broder’s model

Search intent is the goal a person pursues with a search query. Two people can type the same word and want different things. Someone searching for “hall heating” may want to understand which systems exist or may be looking for a provider who can come tomorrow. Search engines try to infer this intent from the query, the location, previous searches and the behavior of many other users.

The foundational model comes from Andrei Broder (2002), then a researcher at the AltaVista search engine. He observed that classic information science understood every search as an information need. On the web, he argued, that is often not the case. He distinguished three types:

• Informational: the searcher wants to know something that is presumably on one or more pages.
• Navigational: the searcher wants to reach a particular website whose address they do not want to type or do not know.
• Transactional: the searcher wants to do something mediated by the web, such as buy, book, download or hire a provider.

For a business, the article on SEO for midsize businesses translates these three types into four search types: knowledge question, comparison question, service plus location, and company name. The knowledge question is informational, the company name navigational, service plus location mostly transactional. The comparison question lies in between: the searcher is informing themselves, but with a purchase intention in the background. In practice it is often called “commercial investigation.”

What do studies show about the distribution of intents?

Jansen, Booth and Spink (2008) automatically classified more than one and a half million search queries from several hundred thousand users of a search engine. More than 80 percent of the queries were informational, about 10 percent each navigational and transactional. To check this, they coded 400 queries by hand. The automatic classification was correct in 74 percent of cases. For the rest, the intent was vague or ambiguous.

These numbers are old and come from a time before smartphones and AI answers. But they show a pattern that still holds: most searches are questions; few are purchase actions. For a business this does not mean it mainly needs knowledge articles. As the worked example in the article on SEO for midsize businesses shows, searches close to the decision bring far more inquiries per visitor. The volume is at the top of the funnel, the order at the bottom.

That search engines show different results depending on intent was examined by Miranda Santos and Teixeira Lopes (2026) on the results pages of major search engines. For informational queries, data-rich elements predominated; for navigational queries, orientation elements including local results; for transactional queries, commercial and decision-supporting elements. AI summaries appeared particularly for queries where someone wants to get something done or learn something.

Two readings: are three types still enough?

The first reading considers Broder’s classification sufficient. It is simple, well researched and can be applied to any query. The second reading says search has changed. Lichtenegger, Urman and Hannak (2026) argue that search engines have turned from link directories into answer providers and that the classification from the early days of the web therefore needs rethinking.

An observational study by Garlough-Shah and colleagues (2026) provides an empirical hint. 52 participants from an online survey had their search behavior recorded between October 2023 and November 2024, 848 search episodes in total. The researchers divided them into information searches (429), purchase-related searches (253) and advice-seeking searches (166). Searches in AI-powered search engines were far longer, averaging 11.6 words, than in traditional search engines with fewer than five words, and AI search engines were used comparatively often for advice-seeking. The sample is small and concerns private individuals. Still, the direction is clear: alongside knowledge, navigation and purchase, asking for a recommendation is gaining weight.

I consider the two readings compatible. Broder’s three types remain the framework. For businesses, however, it pays to consider advice-seeking explicitly, meaning questions like “Which provider of hall heating would you recommend in the region?” How websites appear in such AI answers is described in the article on visibility in AI answers.

The results page test: how to recognize intent without a tool

You do not have to guess the intent behind a query. The search engine has already inferred it from the behavior of many users and shows it in its results. I call the procedure the results page test. It takes five minutes per search term.

1. Run the search

   Type the query the way a customer would, ideally in a private browser window and with your location.

2. Sort the first ten results

   For each hit, note the kind of page: guide, comparison, provider’s service page, directory, map result, video, shop.

3. Note the special elements

   A map box with businesses, question boxes, an AI summary or ads show what the search engine expects for this intent.

4. Determine the dominant kind

   If six of ten results are guides, the intent is informational. A service page will struggle there, however good it is.

5. Assign your own page

   Build or revise the page that matches this kind. If none of your pages fits, the query is a topic for a new page or for none.

Which page fits which search query?

The mapping follows from the test. The table links the four search types to the right page and the elements that must not be missing on it.

For searches with a local reference, the search engine usually shows a map box above the results. There, the business profile has a say, not the website alone. How to maintain it is covered in the article on the Google Business Profile, and the wider context for businesses with a fixed service area is described in the article on local search engine optimization.

For transactional queries, what happens after the click counts. A service page has to make the route to an inquiry short. The structure of such a page is covered in the article on building a landing page.

Which mistakes arise when intent is misread?

Most visibility problems of small websites can be traced back to three mistakes. All three arise because the page is derived from the search term rather than from the intent.

The second mistake is called cannibalization in SEO jargon: two of your own pages compete for the same query, and the search engine cannot decide on either. This contradicts the question-page principle from the article on SEO for midsize businesses: one question, one page, the best answer.

One limit applies to all these rules. Search engines change their results pages constantly, and the intent behind a query can shift, for instance when a topic is in the news. The results page test is therefore not a one-off exercise. For the ten most important search terms it is worth repeating every six months.

How to get started

• Write down the ten search terms through which you would expect the most orders.
• Run the results page test for each term and note the dominant kind of page.
• Assign exactly one of your pages to each term. Mark terms without a matching page.
• Check whether two of your pages serve the same intent and merge one of them.
• Check on every service page whether service area, process, references and contact route are visible.
• Link every expert article to the service it belongs to.

The marked terms without a page are candidates for the next topic plan. How to score them and put them into a quarterly plan is described in the article on the editorial calendar for expert articles. Whether the revised pages then bring more inquiries is shown by the measurement in the article on measuring the website inquiry rate.

Conclusion: which page fits which search query?

We set out to ask
to recognize the intent behind a search and which page fits it. The answer begins with Broder’s classification into informational, navigational and transactional searches. For businesses, the finer form with four search types is more useful: knowledge question to expert article, comparison question to comparison with recommendation, service plus location to service page, company name to homepage, references and profile.

You do not have to guess the intent. The results page shows what the search engine expects for a query. Newer studies suggest that advice-seeking is becoming more important with AI search. That does not change the basic rule; it extends it. If you have exactly one good page for every important intent, you need no tricks. You only need the right page in the right place.