
Keyword density is the percentage of a page's words that are a specific keyword — say, 12 occurrences in 600 words is 2%. Here's the part most guides bury: no modern search engine uses a density target, and optimizing toward one is optimizing for a ranking system that died two decades ago. The real stake today is downside-only: high density can hurt you, but no density helps you.
A zombie metric: why it refuses to die
Keyword density is SEO's most persistent zombie. It gets killed by every Google statement and every ranking study, then shambles back because it's measurable. You can't put "satisfies intent" in a spreadsheet cell, but 2.3% fits beautifully. Tools keep displaying it, so clients keep asking about it, so agencies keep reporting it. The number survives on dashboard-friendliness alone.
The honest history looks like this:
| Era | What SEOs believed | What was actually true |
|---|---|---|
| Late 1990s | Repeat the keyword more than the competition and you outrank them | Largely true — early engines leaned on raw term frequency, and stuffing worked embarrassingly well |
| Early–mid 2000s | There's a magic percentage (2%? 5%?) that maximizes relevance without tripping filters | Engines were already using link signals and TF-IDF-style weighting, where extra repetitions hit hard diminishing returns; the "ideal percentage" was folklore even then |
| 2011–2013 | Keep density "safe" to avoid Panda/Penguin | Quality systems targeted thin and spammy content patterns broadly; density was at most a symptom being measured, never the dial |
| 2013–2019 | Density still matters a little, plus synonyms ("LSI keywords") | Hummingbird and then RankBrain moved matching toward meaning; "LSI keywords" was never a real Google technology, just density-thinking wearing a lab coat |
| 2019–today | — | BERT-class language models evaluate text the way a reader does: context, entities, completeness. A page can rank while using the exact query phrase once, or never |
Google's own people have said versions of "keyword density is not a thing" for well over a decade. The direction of modern relevance is entity- and topic-based — the practical replacement for density thinking is laid out in semantic SEO and NLP.
What the metric is still good for
One thing, honestly: outlier detection. As a diagnostic ceiling rather than a target, density has a use. If a 700-word page uses "best waterproof hiking boots" fourteen times, you don't need a linguistics degree to know the page reads like hell — and a density readout catches that at scale when you can't manually read 500 pages. A tool like this site's keyword density analyzer is useful in exactly that mode: not "hit 2%," but "flag whatever looks compulsive." Past the flagging threshold you're no longer discussing density; you're discussing keyword stuffing, which is a named spam tactic with real consequences and its own glossary entry.
How to check it on your own site
To be clear about the goal: you're screening for excess, not tuning toward a number.
- Crawl your site with Screaming Frog and export word counts. Short pages with aggressive keyword titles are where compulsive repetition usually hides — thin content plus repetition is the classic combo.
- Read your money pages aloud (or have someone else). Anywhere you stumble over a phrase repeated where a pronoun belongs — "our Denver plumbing services" four times in a paragraph — is a rewrite candidate. Your ear is a better parser than any percentage.
- Check old high-traffic pages first. Content written in 2012 under different rules often still carries density-era tics. Sort GSC pages by impressions and review the oldest publish dates.
- Look at anchor text and template zones too. Density thinking often survives in footers, sidebars, and internal anchors ("cheap flights" × 30 in a footer) rather than in body copy.
- After rewriting, compare periods in GSC per URL. Pages purged of robotic repetition typically hold or gain, because the repetition was crowding out the natural vocabulary that actually builds topical relevance.
Common audit mistakes
- Prescribing a target percentage. Any audit that says "increase keyword density to 1.5%" was written by a template. Fix: strike the recommendation; replace with coverage and intent analysis.
- Counting only exact-match occurrences. Repetition of close variants ("SEO agency," "agency for SEO," "SEO agencies") reads just as compulsively. Fix: evaluate the pattern, not one string.
- Diluting instead of rewriting. Teams "fix" high density by padding with filler words, which keeps the ratio pretty and the page bad. Fix: rewrite the repetitive sentences; delete padding.
- Ignoring where the repetitions sit. Ten mentions across a genuine 4,000-word reference are fine; ten in the first 200 words are not. Fix: assess distribution and readability, not the site-wide ratio.
- Letting the tool define the workflow. Because density is what the tool shows, density becomes what the audit optimizes. Fix: demote it to one screening signal among many.
FAQ
What is the ideal keyword density for SEO?
There isn't one. No modern engine scores pages against a density target, and Google has said so repeatedly. Use the keyword where a reader needs it — title, opening, headings where natural — and stop counting.
So the keyword doesn't need to appear at all?
Using the term naturally still helps machines and readers confirm the page's subject — once in the title and early in the body is sensible. That's placement, not density; the tenth repetition adds nothing the second didn't.
Why do SEO tools still show density scores?
Because it's cheap to compute and customers expect it. Treat it as a smoke detector for excess, never as a dial to turn up.
Is TF-IDF the modern version of keyword density?
TF-IDF is a real information-retrieval weighting concept, but the "TF-IDF optimization" sold in content tools is mostly term-suggestion — useful as a coverage hint, not a repetition quota. Modern engines have moved well past bag-of-words scoring either way.
At what point does density become a penalty risk?
There's no published threshold, and that's the point — spam systems detect the pattern of writing for engines instead of people. If repetition is noticeable to a reader, you've crossed the only line that matters. What happens after that line is covered under keyword stuffing.
Claude Vincent is a technical SEO consultant focused on crawlability, rendering, and AI-search visibility. He writes the field guides and case studies at SEO ProCheck, with a bias toward the durable, unglamorous work that decides whether search engines and AI answer engines can actually read and cite a site.
About SEO ProCheck
Technical SEO consulting and GEO strategy with 20 years of enterprise experience. Case studies, resources, and tools for search and AI visibility.
Work With Me
Technical SEO audits, GEO strategy, site migrations, and international SEO. Hourly consulting for teams who need hands-on support, not just reports.







