Image tools

Keyword Density Checker

Paste your text to see the words and phrases you actually repeat, with counts, percentages and a warning when one is overused.

Runs in your browser. Your image is never uploaded.

How to use Keyword Density Checker — Free Online SEO Tool

  1. 1Paste the page text you want to analyse.
  2. 2Review the one, two and three-word phrases with their counts and percentages.
  3. 3Copy the result straight into your page or your CMS.

About Keyword Density Checker

Keyword density started as a ranking lever and ended up as a diagnostic. Nobody optimises to a percentage any more, but the counts are still genuinely useful: they tell you what a page is about in its own words, which is not always what you intended when you started writing. A draft you believe is about "noise-cancelling headphones" and which actually repeats "sound quality" twelve times is worth knowing about before you publish.

The counting is straightforward and deliberately transparent. Text is split on Unicode letters and digits, so accented and non-Latin words are counted properly, then grouped into one-, two- and three-word phrases. Each phrase's density is its share of the total word count, which is why a three-word phrase used four times in 500 words shows a small figure even though it is a strong signal. Common words are skipped by default, with the exception that a common word sitting between two real terms is kept, so "return on investment" survives. Two- and three-word lists are limited to phrases used more than once, and each list shows the top 25.

The 4% flag is a prompt, not a verdict. No search engine has published a density threshold, and the tool says so where it warns you. What usually matters more is the shape of the two- and three-word lists: if the phrases you want the page to rank for are not in them at all, the page is not really about those things, and if one exact phrase dominates all three lists, the copy probably reads repetitively to a human as well. Because everything runs locally, you can paste an unpublished draft or client copy without it leaving your device.

Frequently asked questions

What is a good keyword density?
No search engine publishes a target, and anyone quoting an exact ideal figure has made it up. In practice, naturally written pages usually land somewhere under about 2.5% for their main term. This tool flags single words above 4% as worth a second look — an editing prompt, not a penalty threshold.
How is the percentage worked out?
Uses divided by the total word count, times 100 — so a term used 5 times in a 200-word page is 2.5%. Two- and three-word phrases are measured against the same total, which is the convention SEO tools follow; it is why longer phrases always show a lower percentage than the individual words inside them.
Why are some two-word phrases missing?
Two- and three-word rows only appear when the phrase is used more than once, since a phrase used once is not a pattern. Each of the three lists is capped at the top 25 phrases so the table stays readable, and you can copy or download the full visible table as CSV.
Which words are ignored, and can I include them?
A built-in list of 153 common English words — articles, pronouns, auxiliaries, prepositions and 21 contractions. Untick the option to count them. Phrases where a common word links two real terms, such as "return on investment", are always kept, because dropping them would hide genuine key phrases.
Can keyword stuffing get a page penalised?
Keyword stuffing is named directly in Google's spam policies, so yes, repetition can hurt a page. The trigger is not a density number though — it is text that reads as written for a crawler. If a phrase is over 4% here, read the sentences around it; usually the fix is varying the wording rather than hitting a target figure.
Can it check a live URL instead of pasted text?
No, and that is a real limitation. The whole tool runs in your browser with no server, so it cannot fetch another site. Paste the copy a visitor will actually read — leaving out navigation and footers, which otherwise inflate the counts with boilerplate.