What the checker measures
Keyword density is a ratio: density % = occurrences of a phrase ÷ total words × 100. Paste your text, pick a phrase length of one, two or three words, and the table lists the 20 most frequent phrases with their count and density. The text is lowercased and split into tokens made of letters and digits in any script, with apostrophes kept inside a word, so "Coffee" and "coffee" are counted together, "café" and "don't" stay whole, and hyphenated words such as "well-known" are counted as two tokens.
With one-word phrases and "Ignore common words" ticked, the tool hides a list of 25 English function words (the, a, an, and, or, but, of, to, in, on, at, for, with, is, are, was, were, be, been, it, this, that, as, by, from) from the list. The denominator is always the full word count, with common words included, so a keyword's density is the same whether the box is ticked or not and "Total words" does not change. For two- and three-word phrases nothing is filtered, so phrases like "matter a" appear. The "Unique phrases" figure counts every distinct phrase in the text (after hiding common words), not just the 20 rows shown.
Worked example
Take: "Coffee grinders matter. A good coffee grinder makes better coffee, and burr grinders beat blade grinders. The coffee you brew depends on the grinder."
| Setting | Total words | Top results |
|---|---|---|
| 1 word, common words ignored | 24 | coffee 4 (16.67%), grinders 3 (12.50%), grinder 2 (8.33%); 13 unique phrases |
| 1 word, common words kept | 24 | coffee 4 (16.67%), grinders 3 (12.50%), grinder 2 and "the" 2 (8.33% each); 17 unique phrases |
| 2 words | 24 | every pair appears once (4.17% each) |
Notice that "grinder" and "grinders" are counted separately: the tool matches exact tokens and does no stemming. Add the two counts yourself if you want the combined weight of the term.
How to read the numbers
- It is a diagnostic, not a target. Google does not rank pages by hitting a keyword density, and no percentage is "safe". Google's spam policies do name keyword stuffing, meaning repeating terms unnaturally, so the useful check is whether one phrase dominates the list in a way that reads awkwardly aloud.
- Look for what is missing. If the top phrases are generic and your topic's key terms do not appear at all, the page may not cover the subject clearly. The fix is to write about the subject more fully, not to repeat a word.
- Use longer phrases. Two- and three-word results reveal whether you keep repeating the same exact phrase, which one-word counts hide.
- Language limits. Words in any script are counted, but the common-word list is English only, and scripts written without spaces (such as Japanese) are treated as one long token per run of characters.
- The table shows the top 20. When the text has more distinct phrases, a note says "Showing the top 20 of N", and the "Unique phrases" figure is the full N.
For a plain frequency list without density framing, try the word frequency counter. To check how your title and description present a topic in results, use the SERP snippet checker, and see meta description best practices for writing them naturally.