How many blog posts do I need to rank?

The question gets asked constantly, usually as “is thirty enough” or “do I need a hundred”, and it contains an assumption worth examining before answering: that volume is the input and ranking is the output.

It is not. Ranking is decided per query. A site with one page can rank first if that page is the best answer to a question. A site with four hundred can rank for nothing if all four hundred half-answer the same six questions.

Which means the honest answer is: you need one good page per query you intend to win, and the number of queries you can realistically win is smaller than you think.

Why the question misleads

The volume framing survives because it converts an uncertain creative problem into a countable one. “Publish two a week” is a plan you can execute and report on. “Own the eleven questions our buyers actually ask” requires judgement about which eleven, and offers no way to demonstrate progress in week two.

So teams adopt volume targets, and volume targets have a specific failure mode. Once the calendar requires a post and no genuinely new question remains, the next post is necessarily an angle on a question you already covered. Do that for a year and you have thirty pages orbiting a dozen questions, each one splitting the signals of the others.

The site now has more content and less visibility, and the response is almost always to publish more, because the diagnosis says the content programme is underperforming.

What actually determines coverage

If not volume, then what?

The number of distinct intents you cover well. Not topics — intents. “What is X”, “X vs Y”, “how to do X when Z” and “is X worth it” are four intents that need four pages. But “what is X”, “X explained” and “X definition” are one intent that needs one.

Most sites have far fewer genuine intents available than their content calendar assumes. A specialist B2B product might have thirty to sixty that matter. A local service business might have fifteen. Once covered, publishing more means covering them again.

Depth on a subject rather than spread across subjects. Pages on a related subject reinforce each other — through internal linking, through the accumulated signal that this site handles this area, through the adjacency that lets a page rank for queries it never targeted. Fifteen pages exhausting one subject build something. Fifteen pages across fifteen subjects build fifteen orphans.

Whether each page is the best answer available. The competition is not your other posts. It is the ten results currently ranking. A page that is worse than all ten does not contribute at any volume.

The point where more makes it worse

Worth being precise about, since it is where the real damage happens.

Additional posts stop helping and start hurting when they trigger either of two effects.

Cannibalisation. Two or more pages targeting one intent. Search engines pick one, split signals across the rest, and often rotate which they show, so none establishes a stable position. This is the most common self-inflicted problem on sites publishing regularly for over two years, and the fix is consolidation rather than more publishing.

Quality dilution. Quality assessments operate partly at site level. A domain where most pages are thin is treated differently from one where most are strong, which means adding weak pages imposes a cost on the strong ones. It also spreads crawl attention: a crawler with a finite budget spends it on pages that do not deserve it, so genuinely important pages get revisited less often.

Both are diagnosable. Search your domain for each important query and count how many of your pages appear; check Search Console for the share of your pages receiving zero impressions over six months. A large zero-impression share means you have been paying to make the site harder to rank.

Rough numbers, honestly caveated

Useful as an order of magnitude, not a target.

Two observations. First, for most businesses the number is smaller than the industry implies — under sixty pages covers the entire question surface of a specialist product. Second, in three of five rows the binding constraint is not blog content at all, which is worth knowing before committing a year to a publishing calendar.

A publishing pattern that compounds

Instead of a frequency, a rule: publish when you have a question worth owning, and spend the rest of the time improving what exists.

In practice, for most sites past their first year, that means something close to a 1:2 ratio — one new page for every two substantive improvements to existing ones. The improvements are where the measurable results come from; the new pages are where future results come from.

What improvement means, specifically: take pages at positions 8–20, compare each against the results above it, and close the gap. Usually the gap is that the competing page answers a slightly different question, covers a case yours omits, or is simply current where yours is not. That is a two-hour job with a result visible in weeks, against a new page’s two-hour job with a result visible in months, if at all. The triage is covered in choosing which pages to improve.

Then, once a year, cut. Pages with no impressions after two full quarters should be consolidated into something that works or removed. Deleting content feels like destroying an asset; carrying dead pages is a cost, and removing them regularly produces measurable improvement in what remains.

If you already have 200 posts

Most people asking this question are not starting from zero. They have published for years, traffic has plateaued or declined, and the instinct is that the answer is more.

It is almost never more. Here is the triage, in order.

1. Find the dead weight. Export your pages from Search Console with impressions over the last six months. Sort ascending. Everything with zero or near-zero impressions is a candidate — not automatically for deletion, but for a decision. On a typical two-hundred-page site this is frequently sixty to a hundred pages, which is usually a shock.

2. Sort the dead weight into three piles. Pages that duplicate a better page get merged into it and redirected. Pages covering a real question badly get rewritten rather than removed. Pages covering nothing anyone searches for get deleted and redirected to the nearest relevant page. The third pile is larger than anyone expects and removing it is genuinely beneficial rather than merely tidy.

3. Find the near-misses. Filter to pages at average position eight to twenty. These are your entire near-term opportunity, and on a mature site there are usually more than you can work through in a quarter. Every hour here returns more than an hour spent writing.

4. Only then consider new pages, and only for questions that genuinely have no page yet.

Sites that run this sequence commonly end up with fewer pages and more traffic within two quarters, which is counterintuitive enough that it is worth stating plainly: on a mature site, deletion is a growth tactic. The signals concentrated into the pages that remain outweigh whatever the removed pages were contributing, which for zero-impression pages was nothing at all.

What generation changed

The volume question became more urgent when producing a competent 2,000-word page dropped to minutes, because the constraint that used to limit publishing quietly disappeared.

What did not disappear is everything above. Cannibalisation does not care how the page was written. Quality dilution does not care. Crawl budget does not care. The result is that generation made it dramatically cheaper to cause the specific harms volume causes, which is why sites that scaled publishing aggressively often saw visibility fall.

The failure is not that the content was generated. It is that generation removed the friction that used to force prioritisation. When a page cost a day, nobody wrote the fourth angle on a question they had already covered. When it costs four minutes, everybody does.

So the useful discipline is unchanged and now has to be imposed deliberately: one intent per page, no page published without a specific query it intends to own, and no page published that is not better than what currently ranks for it. Applied to generated content, that produces a real advantage. Skipped, it produces a large site with declining visibility and an unclear cause. The line is covered in AI content that holds up.

One more thing volume does not buy. Assistants composing an answer extract the passage that best addresses the question — they are not counting how many pages your site has. A site with twelve precise, well-structured pages is cited more often than one with four hundred vague ones, because extraction operates on passages and vagueness is not extractable at any scale.

Which returns the question to where it started. Not “how many posts do I need”, but “which questions do I intend to own, and is each of my pages the best available answer to one of them?” That version is harder to plan around and impossible to hit by accident, and it is the only one that predicts whether any of it works.

Frequently asked questions