Keyword Density Checker

Paste plain text or HTML. Before counting, the checker removes script and style blocks, HTML comments and tags, and decodes entities: numeric codes such as ’, and named ones such as &,  , ’ and –. Then it lowercases the text and splits it into words. Anything that is not a letter or a number ends a word, except an apostrophe or hyphen inside one, so “don’t” and “long-tail” each count as a single word.

The results update as you type, and everything runs in your browser.

Density is a count divided by the total number of words, times 100, rounded to one decimal. The total always includes every word, stop words too, whatever the checkbox says. What differs is the count.

So one phrase can show two percentages. The phrase tables say how often it occurs. The focus figure says how much of the text it fills. Neither is wrong; just compare like with like.

Focus keywords must match whole words: “seo” does not match inside “seoul”. Capitals do not matter. Phrases are counted in overlapping runs of words, and punctuation does not break a run, so the end of one sentence and the start of the next can form a phrase.

Google publishes no target or ideal keyword density, as of October 2026. No percentage appears in its documentation for site owners. What it does publish points the other way: its spam policies name keyword stuffing, and one of the examples given is repeating the same words or phrases so often that it sounds unnatural. The policy gives no number for when that starts. It is judged by how the text reads.

Search has also moved past matching strings. Google describes systems that understand synonyms and the meaning of a query, so a page does not need to repeat an exact phrase to be relevant to it. A page that says “water heater” once and then talks plainly about tanks, pilot lights and sediment is about water heaters.

Targets you may have seen come from tools, not from Google. Yoast SEO’s keyphrase density check, for example, turns green between 0.5% and 3%, according to Yoast’s own documentation as of October 2026. That is a rule of thumb that plugin applies, not a threshold Google uses.

This checker has one threshold of its own, and it is about reading, not ranking: a notice when any single word reaches 6% of the text, which is about one word in sixteen. At that rate most readers notice the repetition.

The useful output is not the percentage. It is the list of what you repeat, which is hard to see in your own writing.

Three notices can appear under the text summary. None of them means the page is penalized, and having no notices does not mean the copy is good.

“Only 54 words. Density percentages are unstable on short text — a single extra mention moves them sharply.” This appears under 300 words. On short text the numbers swing: in 54 words, a single mention of a four-word phrase is already 7.4%. Read the tables for repeats and ignore the percentages.

“Repeated heavily: “cheap” at 11.9%…” This appears when any word in the Single words table is at 6% or more, and it asks you to read those sentences aloud. It looks at single words only, so a phrase can repeat often enough to grate while each of its words stays under 6%; check the phrase tables yourself.

“Not found in the text: “best seo plugin”.” This appears for each focus keyword that never occurs as a whole phrase. Before adding it, ask whether the page should target that wording at all.

Ignore common stop words is on by default. The list is short on purpose, 47 English words: articles, pronouns, forms of “be” and “have”, and common prepositions and conjunctions such as the, and, of, to, in, with, you and your. A long list would hide the repetition the tool exists to show.

With the box ticked, single stop words drop out of the Single words table, and phrases made only of stop words, such as “of the” or “in the”, drop out of the phrase tables. A phrase with one real word stays: “the plugin” and “flights to madrid” are still counted. Totals and percentages do not change, only which rows appear.

Untick it to check the small words. A sales page that says “you” in every line, or an About page full of “we”, shows up. Expect function words such as “the” to cross 6% in ordinary writing, so a Repeated heavily notice about them means nothing.

The list is English only. In other languages, the small words are counted like any other and fill the top of the tables; read past them.

An illustrative travel page, run through the checker with the focus keyword “cheap flights to madrid”. The figures below are the tool’s real output for these two texts.

The unique word count tells the story: 33 different words out of 59 against 49 out of 54. The rewrite says the phrase once and spends the rest on what a visitor wants to know.

Two details are worth noticing. “Cheapest” does not count toward “cheap”, because the checker matches exact words, not word forms. And even the rewrite’s 7.4% looks high, only because the text is short. On a 1,000-word page, the same single mention would be 0.4%.

Say you paste a 1,200-word comparison of WordPress SEO plugins, with the focus keywords “seo plugin, best seo plugin”. The values are illustrative. Reading time shows 5 minutes.

Single words: “plugin” appears 84 times, 7%, and the Repeated heavily notice names it. Search the draft and you find most uses are “the plugin” opening a sentence. Swap some for the product’s name, some for “it”, and cut the ones where the subject is obvious. At 52 uses, 4.3%, the notice is gone, and the sentences are clearer about which plugin you mean.

Two-word phrases: “seo plugin” appears 30 times, shown as 2.5%. In the Focus keywords table the same phrase shows 5%, because each use counts its two words. Nothing is wrong; that is the difference between the two formulas.

Three-word phrases: “click save changes” appears 9 times, only 0.8%. No percentage would flag it, but the table shows that the same instruction closes every section. Say it once in a short setup section and cut the rest.

Focus keywords: “best seo plugin” shows Not found. If people search that wording and the post really makes a recommendation, say so once, in plain words, where you give it. If the post does not pick a winner, leave it out.

What you paste decides what gets counted. Paste the post body to check the writing. Paste the whole page only when you want to see everything a visitor gets, menus and footer included.

Avoid pasting the page source from your browser when you check the writing. It brings the menus, footer and sidebar in with your copy, and those words crowd the top of the tables. WordPress also turns apostrophes and dashes into codes such as ’ and –; the checker decodes those, so “don’t” still counts as one word.

SEO plugins run their own checks inside the editor. Yoast SEO’s SEO analysis has a keyphrase density check, which reports a count rather than a percentage. Rank Math’s content analysis includes a keyword density test. SEOPress’s content analysis checks your target keywords. Each uses its own matching rules and limits, so expect their numbers to differ from this checker’s and from each other. None is a figure from Google. If a plugin light is red and the copy reads well, leave the copy.

Most surprising counts come from what was pasted or from how words are split.

Start with the sentences the notice names, and read them aloud. Most fixes are small.

Use the specific name instead of the category word. Use “it” where the subject is clear. Merge sentences that restate each other. Cut a repeated call to action down to one, near the end. Keep the exact term where a reader looks for it, in the title, the main heading and the opening lines, and write naturally everywhere else, using the words your visitors use. The queries in Search Console show you those words.

Run the text through again after editing. The aim is not a lower number but tables with no phrase you would be embarrassed to hear read back.

Some text repeats words because it has to. A high count in these cases is not a problem to fix.

And when every table looks calm, that is not a sign the page will rank. Density says nothing about whether the page answers the query better than the other results do.