Keyword density is the share of a text's total word tokens taken up by a single term or by an exact phrase, calculated by dividing that term's count by the total number of recognized tokens and multiplying by 100. In a transparent keyword density checker example, the same denominator counts every recognized word in the pasted text, including any that the report later filters from the visible table. A standard sample sentence such as "Red blue red. RED green blue red-blue." is a useful starting point because it contains mixed casing, repeated words, and a hyphenated form that interacts with the tokenizer in a specific way. The Unicode-aware tokenizer in the Keyword Density Checker treats "Red", "red", and "RED" as the same token after lowercase normalization, and it splits on punctuation, dashes, and whitespace rather than trying to strip them. Working the math by hand before opening any tool exposes exactly which decisions matter: the denominator, the matching rule, and the way the report decides what to display. Once those three pieces are clear, the percentages on screen stop looking like mysterious scores and start reading like plain arithmetic that can be checked against the source text.
How Keyword Density Is Calculated
Every displayed percentage in the report comes from one consistent formula: the term's count divided by the total number of recognized word tokens, multiplied by 100. The denominator is not adjusted for stop words, the length of the phrase, or the number of possible starting positions in the text. Because the tool runs no language-specific stemming or lemmatization, words such as "runs", "running", and "ran" count as three separate tokens rather than collapsing into one root form. Matching is case-insensitive through Unicode lowercase conversion, so capitalization has no effect on counts or on the resulting percentages.
For an exact phrase, the same denominator applies. The checker tokenizes the phrase with the same rules used for the body text, then counts contiguous overlapping matches inside the text. Phrase density equals the number of matches divided by total tokens times 100, not divided by phrase length and not multiplied by phrase length. Keeping one denominator across single words and phrases is the main reason any percentage in the report can be compared directly with another row in the same report.
| Density type | Numerator | Denominator | Multiplier |
|---|---|---|---|
| Single-word density | Count of the term in the text | Total word tokens | × 100 |
| Exact phrase density | Contiguous overlapping match count | Total word tokens | × 100 |
Worked Example: Eight Tokens From One Short Sentence
Take the sentence "Red blue red. RED green blue red-blue." and run it through the checker's tokenizer by hand. The tokenization rule treats letters and digits as token characters, treats punctuation, dashes, and whitespace as separators, and allows a single internal straight or typographic apostrophe to remain inside a token. The string "red-blue" therefore breaks into two tokens: "red" and "blue".
After splitting on punctuation and whitespace, the recognized tokens are, in order: red, blue, red, red, green, blue, red, blue. That is eight tokens in total. Lowercasing normalizes "Red" and "RED" to "red", which leaves the count at four occurrences of "red", three of "blue", and one of "green".
Applying the formula token by token:
- red: 4 ÷ 8 × 100 = 50%
- blue: 3 ÷ 8 × 100 = 37.5%
- green: 1 ÷ 8 × 100 = 12.5%
The three percentages add up to 100% because those three tokens cover every word in the sentence. A transparent report shows each row with the same denominator underneath, so the visible numbers always reconcile against the source text.
How to Reproduce the Example in the Keyword Density Checker
Paste the sentence into the tool's main text area, optionally enter an exact phrase, set the minimum word length control, and read the report. The exact steps for the worked example are:
- Paste "Red blue red. RED green blue red-blue." into the visible-text field.
- Type "red blue" into the optional exact phrase field.
- Leave the minimum word length control at its default value, or set it to 1, so every token appears in the frequency table.
- Select the Check keyword density action.
- Read the totals block, the top-50 frequency rows, and the phrase density line.
The report will show eight total words, three unique words, four occurrences of "red" at 50%, three of "blue" at 37.5%, one of "green" at 12.5%, and a phrase density of 25% for "red blue" because the contiguous sequence appears twice across eight tokens. For longer drafts the same workflow produces a deterministic report you can compare across revisions, and pairing the paste step with a Microsoft Word keyword density walkthrough keeps the visible copy and the analysis in sync.
Reading the Frequency Table: Counts, Display, and Limits
The frequency table shows at most 50 terms to keep the interface readable, while the total and unique counts still cover the complete input. Rows sort by descending count and then alphabetically for deterministic ties, so two revisions of the same content produce the same row order for matching terms. The minimum word length control affects only which single-word rows appear. Total words and the density denominator still include every recognized token, even if the filter hides one-character rows from view. Setting the minimum to two, for instance, removes any single-letter rows but does not pretend those tokens were absent from the source. This distinction prevents a filtered report from quietly inflating the visible percentages.
The analysis is bounded at 500,000 characters and rejects empty or token-free content rather than returning a misleading zero report. The checker deliberately omits a stop-word list, stemming, lemmatization, synonym expansion, language detection, HTML boilerplate removal, and search-volume data, because those features require language- and corpus-specific reference choices the tool does not make on your behalf. Tokenization uses Unicode letter and number properties, so words written beyond basic ASCII remain countable, and an internal straight or typographic apostrophe can stay inside a token to keep forms like "don't" or "l'été" as single units.
Why Percentage Numbers Don't Predict Rankings
Per Google's published spam policies, repeating a phrase to reach a numeric density target can cross into keyword stuffing, which is the practice of loading pages with terms in an attempt to manipulate search rankings. The checker makes no claim that any particular percentage will improve rankings, and the product contract explicitly avoids embedding any ranking-threshold reference data. Density is descriptive evidence, not a prescriptive target. Use the report to notice accidental repetition, missing terminology, or editorial imbalance, then revise for readers.
Search engines evaluate usefulness, intent satisfaction, originality, structure, links, and reputation, so an isolated density number sits alongside many other signals rather than above them. The checker does not fetch a URL, render JavaScript, compare competitors, connect to Search Console, estimate ranking difficulty, or publish content, and it does not send the text to an AI model. Everything runs in the current browser tab, which helps when drafts should not leave the device, although users should still avoid exposing sensitive text on shared machines. For a useful review, paste the final visible copy, enter the exact phrase whose repetition you want to inspect, run the analysis, and read the surrounding sentences rather than acting on percentages alone. Comparing revisions using the same tokenizer and denominator keeps each result comparable to the last.
Related reading: How to Choose a Separator in a Keyword Combiner.