The short version
- Word count is not a ranking factor. Correlations between length and rankings are real and do not imply causation.
- Longer content often performs better because thorough coverage takes words — not because words cause performance.
- The right length is whatever fully answers the question and no more.
- Padding to reach a target is one of the most reliable ways to make a page worse.
Someone correlates word count with rankings, publishes a chart showing top-ranking pages average some number of words, and a generation of content briefs acquire a minimum. The correlation is real. The inference from it is wrong in a specific and expensive way.
Thorough treatment of a substantial topic takes words. A page that genuinely covers a complex subject will be long as a consequence. Adding words to a shallow page does not make it thorough — it makes it a longer shallow page, and readers detect that immediately.
The short answer
Fit for your workflow
As long as the question requires. No minimum, no maximum.
Some questions are fully answered in 300 words. Some need 3,000. A page that answers its question completely and stops is better than one padded to a target, and better than one truncated to fit a house style. If you must have a rule, make it this: nothing in the page should be there because of a word count.
- Definitional or factual query
- Short. Answer it and stop.
- Comparison or buying decision
- Longer — there is genuinely more to cover.
- How-to with real steps
- As long as the steps require.
- Product or service page
- Short to medium. Length is not the job here.
What the correlation actually reflects
Every row resolves to the same thing: depth causes both the length and the performance. Adding length without adding depth captures none of the benefit, and costs you a reader who now has to scroll past padding to reach the answer.
The case for shorter pages
Answer-extraction has quietly strengthened the argument for concision at the top of a page. A self-contained answer near the beginning is more extractable than one buried under 600 words of preamble — see SEO vs AEO. Long is fine; slow to the point is not.
The practical shape that serves both: answer directly and early, then elaborate at whatever length the subject genuinely needs. That is a structural decision rather than a length decision, and it is the one worth making.
How to decide the length of a specific page
Work from the question rather than from a template. List everything a reader with that question genuinely needs to know in order to act, write those things, and stop. If the list is short the page is short, and a short page that completely answers its question can serve the reader better than a long one that buries the answer.
Where you want a sanity check against competitors, read the pages currently ranking and note what they cover that you do not — then add only what is genuinely missing. That is a coverage check, not a length target, and it is the version of competitor analysis that Google’s people-first guidance is compatible with. See coverage vs density.
A useful editing pass, once the draft exists, is to delete rather than add. Cut anything a reader who already knows the basics would skip, anything restating a point made earlier, and anything present only to set up a section. What remains is usually shorter, denser and better — and if the result is 700 words where the brief said 1,500, the brief was wrong rather than the article.
Padding patterns to cut
A history of the topic nobody asked for. Restating the question for two paragraphs before answering it. Definitions of terms your audience already knows. "In this article we will explore…" Every one of these exists to reach a word count, and every one delays the reader from getting what they came for.