A page can be accurate, well edited and still miss, because it answers a question the searcher did not ask. Google does not tell you, for a given query, whether it expects a tutorial, a tool or a comparison. The evidence is on the results page itself, and reading it takes about ten minutes.
Four intents, and what each one needs
SEO practitioners sort queries into four intents. The list is a convention, not a Google taxonomy and not a ranking factor. Its value is the question behind it: what would make this person stop searching? Plenty of queries mix two intents, which is why the inspection further down matters more than the label.
| Intent | What the searcher wants | The page that satisfies it | Example query (illustrative) |
|---|---|---|---|
| Learn | An explanation or a fact | A guide that gives the answer on the first screen, then the detail | what is a soft 404 |
| Do | A task finished now | A tool, a form, a calculator or a copyable snippet | ads.txt validator |
| Find | One specific site or page | That page, easy to reach under its own name | greenlightmysite seo checker |
| Compare | A choice between options | Side-by-side criteria and a stated recommendation | www or non-www for SEO |
Google’s guidance on helpful content points the same way: it asks whether you made the page first for the people who will read it, not for search engines. Naming the intent before you write is the cheapest honest answer to that question, because it forces you to decide what “finished” looks like for the reader.
The ten-minute inspection
You need a browser, a notepad and the exact words someone would type. If the page already exists, use the query you want it to answer.
- Search in a private window, in your readers’ country and language. Results change with location, language and device, so what you see at your own desk is one sample, not the truth.
- Count the formats among the first ten results. Guides, tools, lists, forum threads, official documentation, videos, product pages. If most of them are tools, an excellent 2,000-word essay is still the wrong shape.
- Read what surrounds the ten links. Video carousels, image rows, related questions and the related searches at the bottom, when they appear, show the other ways people phrase the same need.
- Open three results on your phone. For each one, write down what sits on the first screen before any scrolling: the answer, an input box, a table of contents, an ad, a long introduction.
- Separate the shared from the distinctive. What all three pages do is the entry price. What only one of them does is the gap you can fill.
- Write one sentence: “The searcher wants ___ and is finished when ___.” If you cannot complete it, the query is too vague to build a page around, so pick a narrower one.
If the page is already live, open the Performance report in Search Console and read the queries it appears for. Queries you never planned for often show the intent people actually bring to the page, and the data is free. Keep your sentence from step 6 next to that list and see whether they agree.
Put your page next to the results
Build a grid with one row for each thing a searcher judges in the first seconds. Fill the left column from the results, the next one from your page, and write the decision last. The example below is invented: a page called “Our complete ads.txt handbook” that opens with the history of the standard, while the results for its query are mostly validators and lists of errors.
| Dimension | What the results show | The page today | Decision |
|---|---|---|---|
| Format | Mostly validators and short fix lists | A long essay | Link a tool near the top, keep the essay as the explanation |
| First screen | An input box or a one-line answer | Two paragraphs of history | Move the answer up, history to the bottom |
| Depth | One section per error message | A general overview | Add an error-by-error section |
| Date | A recent date, visible | No date | Add the date and update what changed |
| Angle | The same four steps everywhere | The same four steps | Add what only you can show: a measurement, a config, a case |
| Next step | Links to related fixes | A dead end | Add two links to the next logical task |
The grid also guards against the opposite mistake, which is copying the top result. The results tell you the format and what readers expect to find, not what to say. A page that reproduces the structure of the first result adds nothing a reader could not get from it, and the Angle row of the grid exists to force you to name what yours contributes.
Five outcomes are possible for any page you grid: leave it alone, change the title and the opening, add a section, split it in two, or pick another query. We split when two intents need different formats. A page cannot be a form and an essay at once without being poor at one of them, and that is experience, not a rule from Google.
How we applied it to ads.txt
When we planned the ads.txt material for this site, we read the results for the queries a publisher would type and wrote down what the top pages had in common. Two things kept coming back: a validator you can point at your own domain, and a plain explanation of the file format for the moment a line gets flagged. Those are two intents, “do” and “learn”, and they want different pages. So we built the ads.txt checker for the first and wrote a separate explanation, ads.txt explained, for the second. Each links to the other, and neither tries to be the other.
This is a reading of results pages, not a measurement of outcomes. The site is three months old and its traffic is too small for us to say whether the split helps, and we would rather say that than imply a result. If you want to repeat the exercise on your own topic, list the two or three things the results offer, decide whether they are one format or several, and count how many pages you need. The wider argument for why a tool alone is not enough is in our guide on building the editorial layer around tools.
Make the promise visible where it is made
An intent mismatch usually shows in three places, and each takes minutes to fix. The first is the title. Google builds title links from the title element, the main visible title, the h1 and other sources, so the words in your title and your H1 should promise the same format, and our SEO checker has an “H1 and title aligned” check that catches the plain version of this mistake. Title tags and meta descriptions covers the rest of the rewriting rules.
The second is the description. Google says the page content is the main source of snippets and that it sometimes uses the meta description when it describes the page better, so write one that names the format and the outcome, for example “paste your domain, get line-by-line errors”. The snippet preview shows how it will be cut. The third is the first screen: if the results reward an answer or an input box, put one there, and outline the rest of the page so a skimming reader can find their case (heading structure).
Quick answers
Can one page serve two intents?
Sometimes. If the first screen satisfies the main intent and a clearly separated section serves the second, one page can work. When the two intents call for different formats, such as a tool and an essay, we split them into two pages that link to each other, because a page that tries to be both tends to be mediocre at each. That is our rule of thumb, not a Google rule.
How often should I repeat the inspection?
We recommend doing it before you write a page, before you rewrite one that is losing impressions, and otherwise once or twice a year. The results for a query can change, so a page that matched them a year ago deserves a second look.
What if every result is in a format I cannot build?
Choose a narrower version of the query or accept that this one is not for your page. If every top result is a video or a map, a text guide is unlikely to match what searchers want there. Use the inspection to choose queries, not to force a page into them.
