# Keyword density is not a target: what to check instead
> Keyword density has not been a ranking factor for two decades. Its only remaining use is catching accidental over-repetition that reads badly to humans - above roughly 3% for one term. What still matters is placement: the title, H1, first 100 words and descriptive subheadings.
Source: https://rankcert.com/blog/keyword-density-guide
Published: 2026-08-29 · Updated: 2026-09-01
---
Keyword density - the percentage of a page's words that are your target phrase - was a genuine ranking factor in about 2003. Guides still recommend "1–2%", which is a number invented by nobody, cited by everybody, and used by Google at no point in the last two decades.
## Why it died
Early search engines counted term frequency because they had little else. That was trivially gameable, and pages stuffed with a phrase repeated a hundred times ranked.
Everything since has moved away from counting:
- **TF-IDF** weighted terms by how distinctive they were across the corpus, so common repetition stopped helping.
- **Latent semantic approaches** started recognising that "car" and "automobile" mean the same thing.
- **BERT and its successors** read whole sentences in context, so a page can rank for a phrase it never uses verbatim.
Today a page can rank first for a query that appears nowhere on it, because the engine understands the page is about that topic. Density is not part of the calculation.
Free tool: [Keyword Density Checker](https://rankcert.com/tools/keyword-density) - See which words your content actually repeats and at what percentage, with stop words excluded, so you can spot keyword stuffing before publishing.
## What density is still useful for
Exactly one thing: catching accidental over-repetition that reads badly.
If a 900-word page uses "project management software" fourteen times, no algorithm penalises the ratio - but a human reads it as robotic and leaves. That is the actual cost.
**Above roughly 3% for a single term, read the page aloud.** If it sounds like an SEO exercise, it is. The fix is rewriting the sentences, not swapping in synonyms mechanically, which reads worse.
Density checking is also useful in the other direction. If your target phrase does not appear in the top few terms at all, the page is probably not about what you think it is about.
## What actually matters on-page
Placement, not frequency.
**The title tag.** Still the strongest on-page signal. Lead with the phrase.
**The H1.** Should broadly match the title. One per page.
**The first 100 words.** State what the page is about early, in natural language. This is also the passage AI answer engines are most likely to quote.
**Subheadings.** Descriptive H2s covering the sub-questions of the topic. This is how you demonstrate coverage without repeating one phrase.
**Related terms, not repetitions.** A page about JWTs should mention claims, signing, expiry and base64url. That variety is what signals depth - far more than repeating "JWT" twenty times.
**Internal link anchor text.** Descriptive anchors pointing at the page tell search engines what it is about, and internal anchors are entirely under your control.
## The modern version of the question
Instead of "what is my density", ask:
1. Does the title contain the phrase and lead with it?
2. Does the first paragraph answer the query directly?
3. Do the H2s cover the sub-questions someone with this query also has?
4. Does the page use the vocabulary of the topic, not just the one phrase?
5. Would a knowledgeable person say the page is complete?
A page that passes those five ranks regardless of what a density calculator says. A page that fails them ranks nowhere, whatever the percentage.
The only time density is worth checking is right before you publish, as a stuffing smoke test. Treating a target percentage as a goal produces exactly the writing search engines spent twenty years learning to filter out.