Keyword Density Checker
Analyze repeated words and two- or three-word phrases in your text. Review occurrence counts, clearly defined percentages and filtered results, then export the analysis as CSV.
Maximum 100,000 UTF-16 code units. Input is never silently truncated. Paste plain text: this tool does not fetch URLs or extract content from HTML.
Loading the analyzer…
Keyword frequency results
Filters change displayed rows only. They do not remove words from the analyzed text or change either percentage denominator.
| Word or phrase | Occurrences | Density | Position share |
|---|
CSV includes all terms matching the current filters, not only the displayed row limit. Copy visible rows copies the rows currently shown, as tab-separated text. Percentages are rounded for display.
How to check keyword density
1. Paste plain text
Add the exact content you want to analyze, without HTML markup.
2. Choose words or phrases
Select one, two or three words per term, then click Analyze text.
3. Review and export
Filter the frequency table, inspect the percentages and export CSV if useful.
Understand keyword frequency and density
Keyword density describes repeated terms using a defined numerical formula. It can help you notice repetitive wording or compare drafts, but it does not measure usefulness, factual accuracy or search intent. This tool does not assign an SEO score or prescribe a target percentage.
Two different percentages
Density divides a term’s occurrence count by all recognized word tokens. Position share divides it by the number of eligible positions for the selected term length. For single words, the denominators are equal. For multiword phrases, they can differ.
In “red blue red blue”, there are four word tokens and three eligible two-word positions. “red blue” occurs twice: its density is 50%, while its position share is approximately 66.67%. These are occurrence ratios, not the percentage of words occupied by the phrase.
How words and phrase boundaries are counted
Words are Unicode letter-or-number sequences that can contain combining marks and internal apostrophes or hyphens. Tokens are lowercased using the selected language and normalized to Unicode NFC. Accents and Arabic diacritics are not stripped, and curly and straight apostrophes remain distinct.
Phrases do not cross line breaks or these separators: . ! ? … 。 ！ ？ ؛ ; Other punctuation, such as commas and colons, separates tokens but does not stop a phrase window. This simple rule is not linguistic sentence detection: abbreviations and decimal points can create extra boundaries.
Adjacent phrase windows can overlap. In “a a a”, the bigram “a a” occurs twice. Languages written without word spaces may not be segmented into linguistic words by these rules.
Filtering does not rewrite the analysis
The term filter and minimum-count control hide rows without changing the original totals. The optional English common-word filter is a small explicit list used only for single-word English results. It does not remove words before building phrases or invent new adjacency between remaining words.
Review repetition in context
A frequent term may be appropriate in a technical explanation, product description or focused article. Read the surrounding text before deciding to change it. This tool cannot determine whether repetition is manipulative or whether a page will rank.
Google’s keyword-stuffing guidance addresses unnatural repetition used to manipulate rankings. Use the counts to support editing rather than repeatedly inserting terms to reach a percentage.
For broader text statistics, use the Word Counter. To revise repeated wording, use Find and Replace.
Privacy and local analysis
This tool’s code analyzes the pasted text in your browser. It does not send your input to an API or save it in browser storage. Clipboard access happens when you request a copy, and CSV downloads are generated locally.
Your browser, extensions and other scripts on this website have their own behavior, which this tool does not control.
Frequently asked questions
Does this tool analyze a website URL?
No. Paste the plain text you want to analyze. URLs and HTML pasted into the box are treated as input text, not fetched or parsed as a page.
Is there a recommended density target?
This tool does not recommend a target or judge writing quality from a percentage. Review relevance, clarity and natural wording alongside the counts.
Why does another checker show different percentages?
It may tokenize words differently, remove stop words before counting or use another denominator for phrases. This page displays both formulas so you can interpret the figures.
Does hiding common words change phrase counts?
No. That option only hides single-word English rows. It is disabled for two- and three-word phrase analysis.
Does the tool combine related word forms?
No. It applies lowercase and NFC normalization, without stemming, lemmatization or synonym grouping. “design” and “designs” remain distinct terms.
Can I export more than the visible rows?
Yes. Download filtered CSV exports every term matching the current filters. The row-limit control affects the table and clipboard copy only.