How long should a page be? The honest answer

There is no minimum word count for ranking and no ideal length. The question survives because word count is easy to measure and quality is not, so it gets used as a proxy by people who need a number in a brief.

• A definition that takes three hundred words is finished at three hundred words.
• A technical walkthrough with eight steps and their failure cases is two thousand, and it should be.
• A comparison of two tools is as long as the criteria that decide the choice.
• A page padded to two thousand words to hit a target is a worse page than the same content at eight hundred.

Google's position has been consistent: there is no word count that ranks, and pages should be as long as they need to be. Word count as a target is on its list of things that do not matter, alongside keyword density and meta keywords.

What its guidance does ask is whether the page provides substantial, complete or comprehensive coverage of its topic, which is not the same as length. A page can be comprehensive in four hundred words if the topic is small.

The other end matters more. Producing many pages at scale with nothing of their own is what its spam policies call scaled content abuse, and thin pages written to fill a content calendar are the usual route into it.

1. 1

   Open the top five results

   In a clean window, in your country. These are the pages that satisfied this query.

2. 2

   Count their sections, not their words

   Sections are the useful unit. Five pages with seven sections each tell you the reader expects roughly seven things covered.

3. 3

   List what all five cover

   That list is your minimum. Missing one of them is the gap a reader notices.

4. 4

   List what only one covers

   That is where a page earns its place: the case, the example or the table the others skipped.

5. 5

   Write to that shape, then stop

   Cover the common list, add your own section, and finish. The result lands near the length of the competition without you having counted a word.

These are observations, not targets. A page outside its usual range is a prompt to check whether the topic was bigger or smaller than you assumed.

Both produce a page that does not rank, and they look nothing alike.

• Padding: an introduction about why the topic matters, a history section nobody asked for, a summary that repeats the page, three paragraphs of hedging. The reader leaves before the answer and the page never gets read to the end.
• Stopping early: the page answers the question and skips the two things that happen next, because the writer had run out of knowledge rather than out of space. The reader goes back to the results page, which is the signal you least want to send.

Padding is the more common of the two on a content-calendar site. Stopping early is the more common on a founder's own site, because the founder knows the answer and forgets that the reader does not know what comes after it.

Not by the word count. By these, in order.

1. It answers the query but not the question after it. The reader's next move is another search.
2. Every ranking competitor covers something it does not, and that something is not irrelevant.
3. It states a rule with no example, so a reader cannot tell whether their case is covered.
4. It has impressions and almost no clicks after four weeks, and the title is accurate. The snippet is promising less than the neighbours.
5. Support gets the same follow-up question every time the page is sent to someone.

The fix is never to add words. It is to add the missing section, then read the page again to see what became unnecessary.

Length is a symptom of a split, not the reason for one. Split when the page is answering two different intents, not when it crosses a count.

• Two intents: half the page teaches a concept and half compares products. Those are two searches and two pages.
• A section that people arrive at directly, and that answers a query of its own. Give it its own page and link to it.
• A section growing with every edit because it is really a subject. It will outrank the parent page on its own query once it is separate.
• Do not split to make two thin pages out of one adequate one. One page per topic, and when two of yours compete on the same query, merge them instead.