How to match your content to search intent

Most pages that stall at position twelve are not underwritten. They are the wrong shape. Someone searched for a number and got an essay. Someone searched for a comparison and got a sales page. The words matched, the need did not, and no amount of extra paragraphs will fix a format problem.

This article is the practical half. If you want the concepts first, read search intent explained and come back. What follows is the audit we would run on an existing page: what to record from the SERP, what to compare it against, and which of four decisions to make.

The two-minute version

Open the query in a fresh search. Write down what the top three results actually are, not what they say. Then compare your page against that pattern on four axes: format, depth, angle, freshness. Then pick one of four outcomes: keep it, reshape it, retarget it at a different query, or merge it into something else and retire the URL. That is the whole method. The rest of this is how to do each part without fooling yourself.

One warning before anything else. Matching intent does not mean copying the pages that rank. Match the format, beat the substance. If the top three are all comparison tables, you write a comparison table — a better one, with the columns they left out.

The four intents, in one table

You have seen this taxonomy. It earns one table here and no more, because the labels are the easy part and the work is everything after.

The one thing worth adding: intent is not a property of the keyword, it is a property of the SERP. "CRM" looked informational for years and now returns vendor pages. The only reliable source of truth is the result page as it exists today.

Step 1: read the SERP and record it

Do this in a clean browser window, logged out, so your own history is not shaping the results. Search the query exactly as a real person would type it. Then record five things, in writing, before you look at your own page. Writing it down matters, because if you skip it you will pattern-match to whatever conclusion you already wanted.

1. 1Dominant format. Not the topic — the container. Listicle, single-answer explainer, tool or calculator, comparison table, product page, forum thread, video. If seven of the top ten are the same container, that container is the entry fee.
2. 2Dominant angle. Two guides on the same topic can answer different questions. "How to fix X" versus "why X happens" versus "is X worth fixing". Note which one the titles promise.
3. 3Content type by source. Are these vendors, publishers, government or standards bodies, community posts, or individual practitioners? A SERP full of forum threads means people want a real answer from someone who did it, not a brand explainer.
4. 4What the top three answer in their first screen. Scroll to nothing. Read only what is visible without scrolling on a laptop. Is there a direct answer up there? A price? A table? A definition in the first two sentences? This is the single most transferable observation on the list.
5. 5SERP features present. A featured snippet, a People Also Ask block, a video carousel, a local pack, a shopping row. These tell you what format Google has already decided serves this query, and a local pack usually means the query is not winnable with content at all.

That fourth item deserves emphasis. First-screen behaviour is where most losing pages lose. The ranking pages answer the question immediately and then explain; the stalled page introduces the topic, explains why it matters, and delivers the answer under a heading eleven hundred words down. Same information, different order, different outcome.

The signals that tell you it is an intent problem

Before you audit anything, know which pages to audit. Intent mismatch has a fingerprint that looks different from a technical problem or a thin-content problem.

• High impressions, very low click rate. Google is showing your page for the query and people are reading your title and choosing something else. That is a promise mismatch, and it is the cleanest signal there is. Check it in Search Console per-query, not per-page average.
• High bounce with a fast return to search. People arrive, see it is not what they wanted, and go back. If your analytics can show the time before exit, seconds rather than a minute is the tell. There are other causes, so read why your bounce rate is high before blaming intent.
• Positions 8 to 20 that never improve however much you add. This is the most expensive symptom because it invites the wrong fix. You add 600 words, nothing moves, you add 600 more. A page that is the wrong container does not climb by getting longer.
• You rank for the plural or the variant but not the query you targeted. Often the page is genuinely good and simply aimed at the wrong search. That is a retarget, not a rewrite.
• The page ranks and converts nothing. Traffic arriving on a page built for a different stage of the decision. Covered separately in traffic but no customers.

A page sitting at position 30 or below with almost no impressions is usually not an intent problem. It is a relevance, authority or crawl problem, and you should rule out the technical layer first.

Step 2: compare your page on four axes

Now open your page next to your notes and score it honestly on four things. Not a rating out of ten. Just match, partial, or miss.

Depth means coverage of the questions, not word count. The fastest way to measure it: list every H2 across the top three results, deduplicate them, and mark which ones your page answers. If everyone covers cost and you do not mention cost, that is a real gap and no amount of prose elsewhere substitutes for it.

Freshness is query-dependent and people over-apply it. "What is a canonical tag" does not reward recency. "Google core update" does; Google confirmed core updates in both March 2026 and May 2026, and a page on that topic that stops in 2024 reads as abandoned. Judge it by whether the SERP itself is full of recent dates.

Step 3: pick one of four outcomes

Every audited page gets exactly one decision. Not a list of improvements. One decision, because the four outcomes require different amounts of work and mixing them is how pages get half-rewritten and left worse than before.

Two notes on the last row. Retiring a page is a win, not a failure — two pages competing for the same query is keyword cannibalization, and merging them usually beats optimising either. And when the SERP is dominated by a local pack, a shopping row, or Google's own answer box with nothing below it worth clicking, the honest decision is to stop spending on that query entirely.

A worked reshape

Take a plausible case. A web studio has a page targeting "website maintenance cost". It sits at position 14, gets impressions, gets almost no clicks.

Before

• Title: "Website Maintenance Services — Keep Your Site Running Smoothly"
• Format: service page with a hero image, a benefits list, three testimonial slots and a contact form.
• First screen: a headline about reliability, and a button.
• Body: 700 words on why maintenance matters. No numbers anywhere. Pricing described as "packages tailored to your needs".

What the SERP said

• Dominant format: explainer articles containing a price table. Not service pages.
• Dominant angle: what it costs and what changes the number.
• Top three first screens: all three showed a range or a table above the fold. Two showed it in the first paragraph.
• Features: a People Also Ask block asking what maintenance includes, whether it is monthly, and whether it is worth it.

After

• Title: "What website maintenance actually costs, and what changes it"
• Format: explainer, with a table of tiers by scope in the second block, and the pitch moved to the end.
• First screen: the range, stated plainly, then one sentence on the biggest variable.
• Body restructured around the questions the People Also Ask block exposed — what is included, how often, what happens if you skip it, when a rebuild is cheaper than upkeep.
• Same URL. The commercial intent is honoured by a link to website maintenance after the reader has the numbers, not before.

Note what did not happen. The page did not get longer for the sake of it, the keyword was not repeated more often, and nothing was copied from the competitors. The container changed and the order changed. That is what a reshape is.

Doing this at scale without losing the thread

One page at a time is fine for ten pages. Past that, group the queries first, because pages should be built for clusters of queries with one shared intent, not for single keywords. Two queries belong on the same page when the SERPs overlap heavily. When the SERPs look different, they are different pages no matter how similar the words look.

Search Console is where you find candidates cheaply. Sort your queries by impressions, filter to a click rate well under the site's own average, and you have a list of pages Google is already willing to show for something you are not delivering. Search Console (opens in a new tab) will also show which query a page ranks for versus which one you thought it targeted, which is how most retarget decisions get made.

One more thing to resist: the urge to build separate pages for every phrasing. That produces the scaled-content pattern Google's spam policies describe, where many pages exist mainly to catch rankings rather than to help anyone. It is method-agnostic — hand-written thin pages count too. If you want the version of this focused on making the content earn its keep rather than just rank, read content that gets traffic and leads.

Match the format. Beat the substance. Copying the ranking pages only qualifies you to be the eleventh of them.

How long before you know it worked

Reshaping a page is a real change and Google has to recrawl, reassess and reposition it. Expect weeks rather than days for a meaningful move, and do not judge it on the first week — early volatility after a substantial edit is normal and tells you nothing. Track the query's impressions and click rate together. Click rate rising at a flat position means the promise now matches; position rising means the page is being read as a better answer to the query. If neither moves in a couple of months, your format read was probably wrong, and it is worth redoing Step 1 rather than adding more words.

Resist editing the same page again mid-flight. You will not be able to tell which change did what.

Questions people actually ask

Can one page serve two intents?

Sometimes, and it is the most common way to lose a query. Informational plus commercial can coexist if the order respects the SERP: answer the question first, sell second. Informational plus transactional almost never works, because a product page carrying 1,500 words of explanation is a worse product page and still a worse article. When the SERPs for two queries share few of the same results, build two pages.

Should I change the URL when I reshape a page?

No, in almost every case. Keeping the URL keeps whatever links and history the page has, and the slug carries far less weight than the title and the content. Change it only if the existing slug actively contradicts the new angle, and if you do, 301 the old one and update your internal links rather than leaving them to bounce through the redirect.

How do I read the SERP if my page ranks in a different country?

Search from the market you actually serve, or you will audit against the wrong result page. Results differ enough by country that a UK SERP and an Indian SERP for the same commercial query can call for different formats. Search Console's country breakdown tells you where your impressions come from; audit against that market, not your own location.

Does any of this change for AI Overviews and AI Mode?

Not the method. Google has stated there are no additional requirements to appear in AI features and no special optimizations necessary, so there is no separate intent exercise to run. Practically, answering the question clearly and early is the same behaviour that gets a passage quoted. Search Console added generative AI performance reporting in June 2026, with impressions, pages, countries, devices and dates but no click data, so you can see exposure there and not much else.

What if the top results are all huge brands?

Read what the SERP is telling you, which is often that the head term is not the fight to pick. The intent read still works on the narrower variants, where the format expectations are usually looser and a genuinely specific answer wins on substance. That is the same reasoning behind long-tail keywords, and it is a better use of a reshape than pointing it at a query you cannot win.