🔍 Keyword Density Checker

📝 Input mode

HTML mode strips <script>, <style>, all tags, and decodes the common entities before analyzing.

✏️ Content

Analysis runs automatically as you type (debounced 250 ms). Up to 5 MB.

🛑 Stopwords

English mode filters ~120 articles, pronouns, conjunctions, and auxiliaries before counting.

⚙️ Filters

Minimum word length: {{ minLength }} Show top N entries: {{ topN }}

📊 Document statistics

Paste text on the left to see word, sentence, and paragraph counts.

Total words

Unique words

Characters

Characters (no spaces)

Sentences

Paragraphs

Lexical diversity

Avg words / sentence

Avg characters / word

⚠️ Optimization warnings

• {{ w.message }}

Top-frequency tables appear here once you paste content.

No phrases met the current filter thresholds. Try lowering the minimum word length or disabling stopwords.

#	Phrase	Count	Density	Frequency
{{ idx + 1 }}	{{ entry.word }}	{{ entry.count }}	{{ formatDensity(entry.density) }}

⚠️

ℹ️ About Keyword Density Checker

What keyword density tells you

Keyword density is the percentage of times a word or phrase appears in a body of text relative to the total word count. SEO practitioners use it to confirm that an article actually focuses on its target topic, and to catch over-optimization — repeating a keyword so often that the content looks unnatural to both human readers and search-engine quality signals.

This tool counts unigrams (single words), bigrams (two-word phrases), and trigrams (three-word phrases) in pasted text or HTML, then ranks them by frequency and density. It also reports document-level statistics (word count, sentence count, lexical diversity) and surfaces warnings when a single term crosses common over-optimization thresholds (~3–5% density for a single keyword in a long-form article).

How analysis works

Tokenization — text is split on Unicode word boundaries (the matcher accepts letters, digits, and embedded apostrophes), so contractions like don't stay one token. Comparison is case-insensitive.
HTML mode — when enabled, <script> and <style> blocks (with their contents) are dropped, all remaining tags are stripped, and common HTML entities (&, <, >, ", ') are decoded before tokenization.
Stopwords — when set to English, a ~120-entry stopword list (articles, pronouns, common auxiliaries, prepositions, conjunctions) is removed before the density is computed. Disable this for raw frequency counts that include filler.
Minimum word length — tokens shorter than the threshold (default 3) are discarded after stopword removal.
Density — count / total filtered tokens × 100. The denominator is the same across unigram, bigram, and trigram panels so percentages are comparable.
Top N — only the most-frequent N entries (default 20) are returned for each panel, sorted by count descending, then alphabetically.

SEO density guidance

1–2% — natural density for a focused keyword in long-form content.
3–5% — borderline; the keyword starts to dominate the prose. Vary phrasing, use synonyms, and shift to semantic variants of the topic.
> 5% — flagged by this tool as possible over-optimization. Search engines now use embeddings and entity-based topic models rather than raw counts, so stuffing a term repeatedly rarely helps and often hurts.
Repetitive bigrams / trigrams — a phrase that appears 8+ times in a short article (under 300 words) usually reads as boilerplate. The tool surfaces these as too repetitive.
Short text caveat — under ~100 words, density percentages are statistically unreliable. Treat the table as raw counts, not ratios.

Practical use cases

Audit an existing article before publishing — paste the body text and look for any single keyword above 3% density.
Compare a draft against a competitor — paste each, sort by frequency, and check whether the same topic-defining terms appear at roughly similar densities.
Extract topic phrases — the bigram and trigram panels surface the recurring multi-word phrases that define the article's actual subject (often more telling than single words).
Detect AI-generated boilerplate — repetitive trigrams (e.g. "in this article", "it is important to") are a common artifact of low-effort LLM output.
Lexical diversity — a unique-to-total ratio under ~0.4 in a long passage often signals repetitive writing or aggressive optimization.

Output fields

Word count — total non-stopword tokens after minimum-length filter is applied? No — total tokens from raw text (sentence/paragraph stats use the same denominator).
Unique words — distinct tokens (case-insensitive).
Lexical diversity — uniqueWords / wordCount. Higher = more varied vocabulary.
Average words per sentence — total words divided by sentence count. Long-form writing aims for ~15–20.
Average characters per word — useful for spotting jargon-heavy passages.
Density bar — each row's bar is sized relative to the most frequent entry in the current panel, not the global maximum, so visual ranking is preserved across tabs.

Related references

Google Search Essentials — the modern replacement for the old Webmaster Guidelines; explicitly calls out keyword stuffing as a quality problem.
Wikipedia: Keyword density — historical context and the formula used here.
Wikipedia: n-gram — what unigrams, bigrams, and trigrams actually are in NLP / IR terms.

Limits and privacy

Input is capped at 5 MB. Above that, the analyzer refuses to start — density on book-length text is rarely useful.
The same TypeScript module runs in the browser and behind the API endpoint (Module-First pattern). Browser results are byte-identical to the API output.
Analysis runs locally in the browser — text you paste never reaches a server unless you call the API endpoint explicitly.

API Usage Example:

POST /api/tools/seo/keyword-density-checker
Content-Type: application/json

{
    "text": "The quick brown fox jumps over the lazy dog. The fox is quick.",
    "mode": "plain",
    "stopwords": "en",
    "minLength": 3,
    "topN": 20
}