Strategy & research
Search Intent: A Practical Guide to Reading and Matching It
How to read search intent from the results page rather than from a label, the four dimensions that decide what page to build, how to spot an intent mismatch in Search Console, and a worked example from query to page plan.
8 min read
Search intent is what a searcher is trying to get done when they type a query. Matching it decides whether a good page can compete at all: a well-written article will not rank for a query where people want a calculator. This guide explains where the familiar intent labels come from and why they are not enough to plan a page, the four things intent actually decides, a ten-minute method for reading intent from the results page, how to handle queries with mixed or shifting intent, and how to find pages on your own site that miss it.
The labels, and where they come from
Most SEO tools sort queries into informational, navigational, transactional and commercial. The first three come from a 2002 paper by Andrei Broder, A taxonomy of web search; “commercial” was added later by the SEO industry for people comparing options before buying. They are not Google’s categories.
Google’s own vocabulary is in its Search Quality Rater Guidelines, in the section on understanding user intent. The current version, dated 11 September 2025, uses these:
| Rater guidelines | Meaning | Closest SEO label |
|---|---|---|
| Know | Find information on a topic | Informational |
| Know Simple | A specific answer that fits in a sentence or two or a short list | Informational |
| Do | Accomplish a goal: buy, download, use a tool, get something | Transactional, sometimes commercial |
| Website | Reach a specific website or page | Navigational |
| Visit-in-Person | Find a place nearby | Local |
Two cautions. The guidelines are instructions for human raters who evaluate results, not a description of how ranking works. And older articles still list a “Device Action” category; it does not appear in the current version. The guidelines also say the line between Know and Do is blurry and that strict classification is not the point, which is good advice for content teams too.
Why a label is not enough
“Informational” describes a query like “payroll denmark” and a query like “what is gross pay”, and they need completely different pages. To plan a page, you need four things the label does not tell you. We call them the four parts of intent:
| Part | Question it answers | Read it from |
|---|---|---|
| Task | What is the searcher trying to do: learn, decide, act or go somewhere? | The query wording and the kind of pages at the top |
| Format | What shape of page serves that: guide, list, comparison, tool, template, product page, video? | The format of the top five results |
| Angle | For whom, and how deep: beginner or expert, overview or reference, neutral or opinionated? | Titles, introductions and depth of the top results |
| Context | Does it depend on time or place: this year’s figures, local rules, a nearby business? | Dates on the top results, local packs, country-specific results |
A page that gets the task right but the format wrong, such as an article where the results are all templates, is one of the most common mismatches, and one of the easiest to miss. Write intent as one sentence that covers all four: “A small-business owner in Denmark wants to work out an employee’s holiday pay for this year, and expects a calculator with a short explanation.” If you cannot write the sentence, you have not read the results page closely enough yet.
Reading the results page: a ten-minute method
The results page is the best evidence you have of what Google thinks searchers want, because it reflects what satisfies them today. Search the query in the market you target, logged out or in a private window, and fill in this sheet:
| Look at | Record | What it tells you |
|---|---|---|
| Top 5 organic results | Page type and format of each | Format: if four of five are tools, build a tool |
| Who ranks | Brands, publishers, government, forums, retailers | Task: retailers mean Do; government and publishers mean Know |
| Titles and introductions | Beginner, expert, “best”, “vs”, year in title | Angle, and whether freshness matters |
| Dates on the results | How recently the top results were updated | Context: recent dates across the board mean fresh information is expected |
| Features | AI Overview, People Also Ask, video, images, local pack, shopping | What else answers the query, and what clicks are left |
| Results 6 to 10 | Anything different from the top 5 | Secondary interpretations of the query |
An AI Overview at the top is worth noting but does not change the method. Google says there are no special requirements to appear in AI features beyond normal SEO, and clicks from them are counted in Search Console’s Performance report (see AI features and your website). What it tells you is that the simple version of the answer is already on the page, so your page has to offer the depth, tool or example that the summary cannot.
Query modifiers worth knowing
| Modifier | Usually signals | Page that tends to fit |
|---|---|---|
| what is, meaning, definition | Know Simple | A short answer high on a broader page, rarely its own page |
| how to, guide, steps | Know or Do | Step-by-step guide |
| template, checklist, example, calculator | Do | The asset itself, with a short explanation |
| best, top, vs, alternative, review | Commercial | Comparison or review |
| pricing, price, free trial, buy | Do, purchase | Pricing or product page |
| brand or product name, login | Website | The specific page |
| near me, a city name | Visit-in-Person | Location page |
Modifiers are a shortcut, not a substitute for looking. “Payroll template” could mean a payslip template or a payroll schedule; only the results page says which.
Mixed and shifting intent
Many queries mean different things to different people. The rater guidelines describe a query’s dominant interpretation (what most searchers mean), common interpretations (what many mean) and minor ones (what a few mean), and note that meanings can change over time. That gives you a practical rule:
- Build for the dominant interpretation. That is what the top results serve.
- Serve a common interpretation with a section on the same page if it is a natural next step, or a separate page if it needs a different format.
- Ignore minor interpretations unless they are exactly your audience.
Intent also shifts. When the top results for a query you rank for move from explanatory articles to tools or comparisons, an accurate page slides. Recheck the results page for your most valuable queries once or twice a year, and record the date you checked. A shift like that is one of the causes covered in what content decay is and how to detect it.
Finding intent mismatches on your own site
You do not need to search every query by hand. Search Console points you to the pages most likely to be mismatched. These patterns are rules of thumb, so treat them as a shortlist to check, not a verdict:
| Pattern in Search Console | Possible mismatch | Check |
|---|---|---|
| High impressions, low CTR, position in the top 10 | Title promises the wrong format, or a feature above takes the clicks | Results page: format of the top results, and what sits above them |
| Position stuck between about 8 and 15 for the main query | Right topic, wrong format or angle | Compare your format with the top five |
| The page ranks for queries it does not answer | Google sees the page as relevant to a need it does not meet | Add the missing answer, or point those queries at a better page |
| Clicks hold but engagement and key events fall | The visit is not satisfied | Read the page as a searcher; compare with what the top results offer |
Read average position with care: Search Console reports the topmost position your result appeared in, averaged across the queries it appeared for, as described in Google’s definitions. Check the query-level numbers before concluding anything from a page-level average.
Search Console
SerpApiWorked example: from query to page plan
An illustrative example, not client data. A payroll software company wants to rank for “holiday pay denmark”. The results-page sheet shows:
- Top five: the government’s official guidance, two accounting-firm guides, a calculator, and a union’s explainer.
- Titles: “rules”, “how it works”, “calculate”; three of five mention the current year.
- Features: an AI Overview summarizing the basic rule, and People Also Ask questions about part-time staff and leaving employees.
Reading the four parts: the task is mostly Know (understand the rules), with a common Do interpretation (calculate an amount). The format is a guide, with a calculator as a strong second. The angle is practical, for an employer rather than an employee. The context is time-sensitive: figures change each year. The plan that follows:
| Decision | Because |
|---|---|
| A guide for employers, owning “holiday pay denmark” | Dominant interpretation is understanding the rules |
| Link prominently to the existing calculator, rather than building a second one into the guide | The calculator already owns the Do intent; two calculators would compete |
| Sections on part-time staff and leaving employees | The People Also Ask questions and the official guidance’s structure |
| A worked example with this year’s figures and a visible “updated” date | The AI Overview covers the basic rule; the example is what it cannot give |
| Link to the official guidance as the source | Accuracy, and the government page ranks first for a reason |
Where intent goes next
Intent is the input to three other jobs. The keyword-to-page map uses it to decide which queries share a page. The content brief puts the intent sentence at the top so the writer builds the right format. And the content strategy uses it to place pages in the buying journey, so the site has pages for learning, deciding and acting, not only for learning.
Where Morrison fits
Morrison’s site index records a search intent and a page type for every URL, so you can filter a section to, for example, its informational pages and check them against the queries Search Console says they get. Chat connects to Search Console for the queries each page gets and to SerpApi for live Google results, including which features appear, so it can compare a page with what ranks for its main query, as in the film above. To check every page in a section, run a workflow: Search Console, SerpApi and AI steps on up to 500 pages a run. See search intent alignment.
Key takeaways
- The labels are a start: informational, navigational and transactional come from a 2002 paper; Google’s raters use Know, Know Simple, Do, Website and Visit-in-Person.
- Plan with the four parts: task, format, angle and context, written as one sentence.
- Read it from the results page, with a fixed sheet, in the market you target.
- Build for the dominant interpretation; serve common ones deliberately and recheck as intent shifts.
- Use Search Console to shortlist mismatches, then confirm on the results page.
Sources and further reading
- 1Search Quality Rater Guidelines (General Guidelines, September 11, 2025)Google
- 2SEO Starter GuideGoogle Search Central
- 3AI features and your websiteGoogle Search Central
- 4What are impressions, position, and clicks?Search Console Help
- 5Performance report (Search results)Search Console Help
- 6Creating helpful, reliable, people-first contentGoogle Search Central
- 7A taxonomy of web search (Broder, SIGIR Forum 36(2), 2002)ACM
CEO, Morrison
Ulrich is CEO of Morrison and founded Bonzer in 2017, growing it into one of Scandinavia's leading SEO agencies with 900+ clients across Copenhagen, Oslo, and Stockholm. At Morrison he leads strategy, operations and go-to-market, bringing years of hands-on SEO and content work to the platform side of the business.