Keyword density is the percentage of a text made up of a given word or phrase: occurrences divided by total words, times 100. It is simple to compute and widely misunderstood, so it is worth being clear about what it is good for.
It is a diagnostic, not a target. There is no density figure that makes a page rank. Search engines stopped weighting raw term frequency in any straightforward way a very long time ago, and writing to hit a percentage produces text that reads badly and can trip spam detection. What the number is genuinely useful for is telling you what a piece is actually about, which is often not what you intended.
What the number is good for
Catching unintentional repetition. If “solution” appears 34 times in 800 words, that is worth knowing, and it is the kind of thing you stop noticing in your own draft.
Checking the topic is where you think it is. Run a competitor’s page and your own. If theirs surfaces the terms a reader would search for and yours surfaces filler, that is a real signal — not about density, but about focus.
Spotting missing vocabulary. A page about mortgages that never mentions “interest”, “repayment” or “deposit” is probably underspecified. Search engines are looking for topical coverage, and the absence of obvious related terms is more diagnostic than the presence of one repeated term.
Finding accidental keyword cannibalisation. Two pages heavily weighted toward the same phrase will compete with each other.
The old advice, and why it is wrong
For years the guidance was to aim for 1–3%. Ignore it. It comes from an era when search engines counted terms more or less literally, and following it now produces two problems at once: prose that repeats itself unnaturally, and a page that looks optimised rather than written. Modern retrieval works on semantic relationships — synonyms, related concepts, entities and context — so a page can rank strongly for a phrase it never uses verbatim, and a page that uses it forty times can rank for nothing.
The practical replacement: write for the reader, then use density to check you have not drifted. If your target phrase appears naturally a handful of times in a substantial piece, that is fine. If it appears once in 2,000 words, the page may genuinely not be about it.
Stop words
Function words — “the”, “and”, “of”, “to”, “a” — dominate any raw frequency count and tell you nothing. Most useful analysis filters them out. Watch for the case where a stop word is part of your actual phrase (“work from home”), which is a common source of confusion in these reports.
Where people go wrong
Writing to a percentage. Produces stilted prose and can look manipulative. This is the whole trap.
Ignoring phrases in favour of single words. People search in phrases. Two-and three-word sequences tell you far more about a page’s topic than individual terms.
Counting the whole page. Navigation, footer and boilerplate dilute the figure and vary between templates. Analyse the body copy.
Treating high density as automatically bad. A glossary entry for a term will legitimately use that term constantly. Context decides.
Using density as the only measure. It says nothing about whether the content answers the question, which is the thing that actually matters.
Pair this with the readability score to check the prose is not only on topic but comprehensible, and the word counter for length.
Your text is processed entirely in your browser and never transmitted — see the privacy policy.