Word frequency counter — words, phrases and density
Count how often each word or phrase appears in a piece of text, with stop words, minimum counts and two- and three-word phrases — and an honest account of what the numbers do and do not mean.
Runs in your browser — your input is not uploaded to ToolsNow.
Separate with spaces or commas.
The text you enter is processed locally in your browser and is not uploaded to ToolsNow.
How words are counted. A word is a run of letters, digits and marks in any script, so accents and non-Latin writing count properly. Apostrophes and hyphens inside a word are kept, so “don’t” and “well-known” each count as one. Everything else separates words. The share column is that term as a percentage of the tokens actually counted after your filters — so switching stop words off and on changes the denominator, not just the rows.
On keyword density. The percentage here is arithmetic, not a score. Search engines have not used keyword density as a ranking signal for well over a decade, and no threshold exists to aim for — writing to hit a percentage produces worse text and no ranking benefit. Where frequency counts are genuinely useful is in seeing what a document is actually about, spotting a word you have leaned on too heavily, and comparing drafts.
How to use
- Paste your text. An article, an essay, a transcript, a batch of reviews.
- Choose what to count. Single words, or two- and three-word phrases.
- Filter the noise. Exclude common stop words, set a minimum count or word length, and add your own words to ignore.
- Read or export. Search within the table, then copy it or download CSV or JSON.
Questions
What counts as a word?
A run of letters, digits or marks in any script, so accented and non-Latin text is counted properly rather than being split apart. Apostrophes and hyphens inside a word are kept, so "don't" and "well-known" are one word each. Everything else — punctuation, symbols, whitespace — separates words. Stating the rule matters, because tools that tokenise differently give different totals for the same text.
What is the percentage column?
That term as a share of the tokens actually counted, after your filters. Turning stop words off and on changes the denominator as well as the rows, so the same word can show a different percentage under different settings. That is arithmetic working correctly, not an inconsistency.
What keyword density should I aim for?
None. Search engines stopped using keyword density as a ranking signal well over a decade ago, and no target percentage exists — writing to hit one produces worse prose and no ranking benefit. Frequency counts are genuinely useful for seeing what a document is actually about, catching a word you have overused, and comparing drafts. They do not measure quality or relevance.
What are stop words, and should I exclude them?
Grammatical filler — the, of, and, is — which dominate any frequency list and say nothing about the subject. Excluding them surfaces the content words, which is usually what you want. The list here is deliberately conservative: function words only, no topic words, so switching it on cannot quietly hide something meaningful.
What are two- and three-word phrases for?
They show what a text says rather than which words it uses. "Machine" and "learning" appearing often is a weaker signal than "machine learning" appearing as a phrase. Stop words are removed before phrases are built, so excluding them turns "the big cat" into the phrase "big cat" rather than leaving a gap.
Will a long document freeze the page?
No. The analysis runs on a deferred copy of your text, so typing stays responsive while counting catches up, and the table renders the first 500 rows while exports include everything.