How many keywords per page should you target for SEO?

One page should target one primary keyword and the cluster of close variants that share its search intent, which in practice usually means somewhere between three and fifteen related phrasings. The number itself is not the point. The rule that matters is intent: if two keywords would be satisfied by the same page, they belong together on that page, and if they would not, no amount of subheadings will make them fit. Counting keywords is the wrong question. Counting intents is the right one.

Why "how many keywords per page" is the wrong question

The question comes from an older era of SEO, when pages were built around exact-match terms and you genuinely did need one page per keyword because the algorithm was matching strings. That has not been true for a long time. Google has been resolving synonyms, plurals, word order and paraphrase for years, which means a single well-built page routinely ranks for hundreds of queries nobody explicitly targeted.

Here is the useful way to see it. You do not choose how many keywords a page targets. You choose how many questions a page answers, and the keywords follow. A page that thoroughly answers one question will pick up every phrasing of that question automatically. A page that tries to answer four unrelated questions will answer all of them badly and rank for none.

The test that actually decides it

Take two keywords and ask: would a person typing either one be happy landing on the same page? If yes, one page. If no, two pages. That is the whole test, and it is more reliable than any number.

"Keyword gap analysis" and "how to find competitor keywords" pass the test. Somebody typing either is after the same thing, and one page serves both. "Keyword research tool" and "keyword research tutorial" fail it. The first wants a product to evaluate, the second wants to be taught. Same topic, different jobs, and one page cannot do both without frustrating half its visitors.

So what is the practical range?

For most pages, a primary keyword plus roughly three to fifteen secondary variants is the natural size of a real intent cluster. It lands there not because of a rule but because that is genuinely how many ways people phrase a single question. Some clusters are smaller. A narrow commercial term might have two variants and that is fine. Some are larger: a broad pillar page can legitimately carry forty related phrasings because the topic really is that wide.

What should worry you is not the count but the spread. If your list of secondaries contains terms with genuinely different intents, you do not have one page with many keywords. You have several pages wearing a trench coat, and Google will treat them accordingly: the page will rank mediocrely for everything and well for nothing, because it is a compromise between incompatible readers.

How to build the cluster for a page

1. Pick the primary. The highest-volume term that genuinely matches what the page is for. This is the one that leads your title and your H1.
2. Collect the phrasings. Every way people ask that same question: singular and plural, word-order variants, synonyms, the question forms, the "for X" qualifiers.
3. Cut the impostors. Anything that reads similar but wants a different answer. These feel like they belong and they are the reason pages get muddled.
4. Map to structure. The primary goes in the title and H1. The secondaries become H2s, H3s and the questions you answer in the body. If a secondary cannot find a natural home in the outline, it probably failed step three.

That mapping step is where the count answers itself. If you can give every secondary a real section that a reader would want, they belong. If you are inventing a paragraph purely to house a keyword, you have your answer, and the keyword goes on a different page or nowhere at all.

What about keyword density?

Ignore it. There is no target percentage, there has not been one for over a decade, and chasing a number reliably makes writing worse. Use the primary keyword where it would appear naturally: the title, the H1, early in the opening paragraph, and wherever the topic genuinely comes up. Then write for the person reading. If a keyword only fits by force, the sentence was not about that keyword.

The related worry, that mentioning a term too often looks like stuffing, is mostly unfounded when you are writing naturally. Pages about a topic mention the topic. That is not a trick, it is just what writing about something looks like.

The failure mode nobody talks about

Splitting too finely is the more common and more expensive mistake, and it is the one that comes from taking "one keyword per page" literally. Somebody finds twelve variants of the same question in a keyword export, treats each as a target, and publishes twelve thin pages. Now every one of them is weak, they compete with each other, the internal links and any external links are spread across a dozen URLs instead of concentrated on one, and the site has a keyword cannibalization problem that took real effort to build.

One strong page beats twelve thin ones essentially every time. It is not close. The twelve pages divide everything that makes a page rank, and division is the opposite of what you want.

The opposite failure: pages that carry too much

Less common, but real. A single page trying to be the definitive resource on an entire category ends up serving nobody in particular. The signal is a page that ranks on page two for a lot of things and page one for nothing. When that happens, the fix is to split by intent, keep the broad page as a pillar that links down to focused pages beneath it, and let each child own one job. The topic cluster model exists precisely to make that structure deliberate rather than accidental.

Where head terms are a different problem

Everything above is about matching intent, which is what you control. There is a separate constraint worth being honest about: for genuinely competitive head terms, getting the keyword structure right is necessary and not sufficient. Those positions are held by pages with authority behind them, and a perfectly targeted page with nothing pointing at it from other sites will sit at position 14 forever while a worse page outranks it. Structure decides whether you are eligible. It does not decide whether you win.

Which is another argument for the long tail first. Those terms are winnable on relevance alone, and they compound into the authority that makes the head terms reachable later.

The short version

Target one intent per page, which usually looks like one primary keyword and three to fifteen variants, and let the exact number fall where the intent actually sits. Use the "would both searchers be happy here" test to decide. Map the primary to your title and H1, the secondaries to your outline, and drop anything that will not sit naturally. Then stop counting and write the page. The keyword clustering tool does the grouping step automatically if you would rather not do it in a spreadsheet, and the SEO content strategy page shows what the finished set of clusters looks like once it is ordered into something you can publish.