How Do You Write a Page That Answers a Query Instead of Circling It?

Somebody types a question about whether to use one thing or the other. They land on a page whose title repeats their question back to them almost word for word. Nine paragraphs later they still do not have an answer, because the page has explained what both things are, listed considerations for each, and closed with the observation that the right choice depends on your needs.

That page is circling. It is not thin, it is not spam, and it was almost certainly written by somebody who knew the answer. It simply never said it. Search intent content is the discipline of resolving the query where the reader is actually looking, which is the first screen, and then earning everything underneath.

The fix is structural, not stylistic. Circling is rarely caused by weak prose. It is caused by an outline that saves the conclusion for the end, because that is where conclusions go in an essay. A web page is not an essay. Nobody scrolls in order to be persuaded that the author was thorough.

What circling actually looks like

Circling has two components and most underperforming pages have both. One is a delay at the start. The other is a refusal in the middle. Together they produce a page that reads competently and helps nobody, which is why it survives review.

The warm-up

The first two paragraphs restate the reader’s situation back to them in general terms. Buyers have more choice than before. Teams are under pressure. Technology moves quickly. All true, all irrelevant, and all written after the real work was done, because an opening paragraph is the last thing anyone wants to write and the easiest thing to fill with atmosphere.

The hedge

Then, when the answer finally appears, it arrives wearing a coat. It may be worth considering. Many organisations find. Depending on your circumstances. Each qualifier is individually defensible and collectively fatal, because a reader cannot act on a sentence that has been pre-emptively withdrawn. The tells are easy to check against a live page.

• The page defines the term in its own title. If somebody searched for the phrase, they have met it before. A definition is context, not an answer.
• A subheading reads "so what should you do?" That subheading is an admission. Everything above it was preamble the writer knew was preamble.
• Every verdict carries a hedge in the same sentence. Qualifications belong in the paragraph after the claim, where they refine it, rather than inside it where they dissolve it.
• The comparison table has no recommendation under it. A neutral table hands the work back to the reader, who came to you precisely because they did not want to do it.
• You can delete the first three hundred words. If nothing is lost, they were never doing anything. Delete them, and notice that the page now starts better.

Why competent writers circle anyway

The causes are worth understanding, because telling a writer to be more direct fixes nothing if the incentives point the other way. Four causes account for nearly all of it: a brief that specifies a word count instead of a question, a keyword given without the query behind it, an approval chain in which anything definite gets softened by somebody protecting the brand, and plain professional caution about being wrong in public where the page will sit for years.

The last one deserves respect rather than mockery. Being specific in writing is a commitment, and specific pages age in public. But the answer to that risk is to name the conditions under which your recommendation changes, which is more useful than hedging and considerably braver. That is also the practical version of E-E-A-T without the mysticism: experience shows up as the willingness to say which situation you are describing and which one you are not.

Word-count briefs deserve less respect. A target of eighteen hundred words on a question that resolves in four hundred is an instruction to pad, and padding is what circling is made of. Brief the question and the reader instead, then let the length settle where it settles.

Search intent content: four jobs, four first screens

Intent is not a taxonomy exercise. It decides one thing only: what has to be visible before a reader scrolls. Get that wrong and the page fails regardless of how good the rest is.

Read the rows across and the pattern is plain enough: every failure is the same failure wearing different clothes. The page delivers the thing the writer found easiest to produce rather than the thing the query asked for.

The mistake of writing one page for two intents

This is the failure we see most, and it is usually the result of a sensible-sounding decision. Somebody notices that two queries share a phrase, decides one strong page will serve both, and writes an explainer that turns into a sales page halfway down. The result serves neither reader, because a first screen can only make one promise, and this one makes two in a tone that fits neither.

The test is behavioural rather than lexical. Ask what the reader does immediately after they are satisfied. Somebody who wanted to understand a term goes away better informed, and possibly no further. Somebody who wanted to buy asks about scope and price. Those are different people on different days, even when they are the same person, and one page cannot end well for both.

The inverse mistake also exists, and it is why this is a judgement rather than a rule. Two pages that answer the same question in slightly different words compete with each other, split whatever authority they have, and confuse anyone linking to either. When that happens the right move is to merge, which we have written about separately in when to consolidate two pages instead of optimising both. Same intent, one page. Different intent, separate pages that link to each other honestly.

The shape of a page that resolves

Here is the running order we use. It is not a template to fill in; it is an order to think in, and it works because each step answers a question the previous step provoked.

1. The answer, in one or two sentences. Written so it could be quoted on its own and still be correct. If you cannot write it, you are not ready to write the page, and no amount of research below it will disguise that.
2. The condition that changes the answer. One sentence naming who this does not apply to. This is where the honesty a hedge was trying to buy actually gets delivered.
3. The mechanism. Why the answer is true. A reader who understands the mechanism can apply it to a case you never wrote about, which is the difference between a page and a reference.
4. The comparison, with the verdict attached. Tables are excellent and incomplete. Every table needs a sentence underneath saying which row you would pick and in what circumstances.
5. The edge cases, named. Regulated sectors, very small teams, legacy constraints. Naming them stops the page from being wrong for a reader it was never written for.
6. The next action. One thing to do on the same day, described concretely enough that somebody could start it without asking a follow-up question.

A page that answers on the first screen is not a shorter page. It is the same material in a different order, with the verdict moved to the front and the caution moved behind it.

How the rest of the page earns its place

Answering early raises an obvious objection: if the answer is at the top, why would anyone read on? Because an answer creates questions. Having been told what to do, a reader immediately wants to know why it is true, what it costs, what happens in their specific situation, and what the person telling them has actually seen. Those are four sections that write themselves once the answer is fixed.

What does not earn its place: a glossary of terms the audience knows, a history of the discipline, a restatement of the answer in different words, and a closing paragraph that summarises the page for a reader who just read it. Cut all four. The space is better spent on one worked example, which is the single most persuasive block in content and organic growth and the part most often skipped because it is the hardest thing to write.

Structure also pays off in how the page gets reused. An answer written as a self-contained statement is the unit that assistants and search features can lift and attribute, and a hedged one is not. The same logic governs which schema types earn something: markup describes what is already visible and legible, and it cannot rescue a page whose answer is spread across four paragraphs of qualification.

Three tests before it goes live

These take about fifteen minutes together and catch nearly everything, which is a good ratio for a review step that anyone on the team can run without training.

The first is the first-screen test. Read only what is visible before scrolling, out loud, on a phone. If the query is not answered in what you just read, the page circles. The second is the extraction test. Lift the answer paragraph out of the page and read it cold. If it needs the surrounding text to make sense, it is not an answer yet. The third is the deletion test: remove the opening three hundred words and see whether anybody notices.

Running this across an existing library is a different exercise, and the trap there is producing a spreadsheet nobody can act on. Sixty pages marked "needs work" is not a plan. Each page needs a severity and a separate call on priority, because the worst page on the site may also be the one nobody visits. A tool such as Prooflin exists to resolve review notes into that shape, and our parent company has written about reading a report you commissioned in how to read an audit you were sold, which is the companion problem.

If you would rather have somebody else run the pass, that is what an SEO audit should include, and a report that never says which page to rewrite first has the same defect as the pages it is reviewing.

When circling is the correct choice

There is a real exception, and it is not the one people reach for. When the honest answer is that the question is wrong, you cannot answer it directly, and pretending otherwise is worse than hedging. A reader asking which of two tools to buy, when neither solves what they described, needs the question reframed before anything else. Do that in the first screen too: name the assumption, say why it does not hold, then answer the question they should have asked.

The other exception is a query nobody is running yet. If the category has no established language, there is no intent to resolve and the page has a different job: it has to teach a reader to recognise a problem before it can recommend anything. We treat that as a separate discipline in marketing a product nobody is searching for yet, because the usual structure genuinely does not apply.

Everywhere else the rule holds, and it is worth stating plainly enough to argue with: a page earns its scroll by resolving something, not by promising to. Start with the page that gets the most impressions and the fewest clicks, because that is a page where people came, read the promise, and decided it would not pay out. Rewrite the first screen only. Leave the rest untouched for a month so you can see what the first screen did on its own, which is also how you keep any honest read on conversion work. If you want a second pair of eyes on a page that should be working and is not, send it to us with the query it was written for.