Analyze keyword frequency and density in your copy, with repeated-word and phrase tables, stop-word filtering, and JSON or CSV export.
Content last reviewed
Paste a blog draft, landing page, product description, or webpage copy.
The checker will calculate word count, reading time, focus keyword density, repeated words, two-word phrases, three-word phrases, and practical overuse signals.
Drop in the article, landing page, or product copy you want to review.
Add an optional focus phrase and comma-separated secondary keywords, and choose whether stop words appear in the tables.
Click Analyze content to see counts, densities, repeated words and phrases, and guidance — then copy JSON or CSV if you need a record.
A 120-word draft where the phrase cold brew appears four times.
Input
Focus keyword: cold brew · Content: a 120-word product description mentioning the phrase four timesOutput
cold brew — 4 occurrence(s) · 3.3% density · status High: usage is high, check whether repeated mentions still sound naturalNormalization treats hyphens as spaces, so a focus phrase matches its hyphenated form too.
Input
Focus keyword: cold brew · Content includes both "cold brew" and "cold-brew"Output
Both spellings are counted as the same two-word phrase in the totalRepetition is hard to hear in your own writing. A phrase that felt natural in every paragraph reads as keyword stuffing to a visitor — and thin coverage of the topic you meant to target is just as invisible from the inside. This checker measures both: paste your draft, name your focus phrase, and get exact counts, density percentages, and tables of the words and multi-word phrases your text leans on most.
The analysis starts with the basic shape of the text — word, character, sentence, and paragraph counts, unique words, and an estimated reading time — then examines keyword usage. The focus keyword is matched as an exact phrase (multi-word phrases work), counted, and expressed as a percentage of total words with a plain-language status: not found, low, natural, high, or possibly overused. Secondary keywords, entered comma-separated, each get the same treatment. Below that, three tables show the top repeated single words, two-word phrases, and three-word phrases, with common stop words filtered out by default so the tables surface meaningful terms rather than the and of.
Paste your draft into the Page or article content box — an article, landing page, product description, or any copy you are reviewing.
Optionally enter a Focus keyword or phrase (multi-word phrases like keyword density checker are matched exactly) and Secondary keywords, separated by commas.
Decide whether to enable Include stop words in repeated-word tables — off by default, which keeps words like the and to out of the tables.
Click Analyze content. The results show summary statistics, the focus and secondary keyword cards with density and status, and three tables of repeated words, two-word phrases, and three-word phrases.
Revise the text where the guidance flags high or overused terms, then paste the new version and analyze again.
Use Copy JSON for a structured summary or Copy CSV for the keyword table when you want a record for editorial review.
Focus and secondary keywords are counted as exact phrases after normalization, so multi-word terms are measured correctly.
Each keyword shows count, percentage, and a banded status — not found, low, natural (0.5–2.5%), high (2.5–4%), or possibly overused (over 4%).
Top 20 single words plus the top two-word and three-word phrases that appear more than once, revealing the patterns your text leans on.
Common function words are excluded from the tables by default and can be included with one switch.
Word, character, sentence, and paragraph counts, unique words, and an estimated reading time at about 220 words per minute.
Copy a structured summary or a keyword,count,density table for editorial records.
Check that a blog draft uses its focus phrase naturally — present, but not in every paragraph.
Spot phrases repeated enough to read as pushy or spammy in conversion copy.
Compare how a main keyword and its variants are distributed across a long article.
Use the repeated-word table to catch the words you overuse without noticing.
Export the JSON or CSV summary as a snapshot for content review workflows.
When you click Analyze, the text is normalized and tokenized: lowercased, accents and apostrophes stripped, hyphens, underscores, and slashes treated as spaces, and remaining punctuation removed. This is why cold brew and cold-brew count as the same phrase, and why matching is case-insensitive.
Keyword counting slides the focus phrase across the token stream looking for exact consecutive matches, then divides by total words for the density percentage. The status bands are: not found (0), low (below 0.5%), natural (0.5–2.5%), high (2.5–4%), and possibly overused (above 4%) — presented as review guidance, not targets.
The tables count individual words (skipping stop words unless you include them) and two- and three-word sequences, keeping phrases that occur more than once. With stop words excluded, a phrase must still contain enough meaningful words to be listed, which keeps fragments like "of the" out of the phrase tables. Analysis runs when you click the button, so you can edit the input freely without the results shifting under you.
Pasted article or page copy. Paste rendered text rather than raw HTML — markup tags would be counted as words after normalization.
A structured summary (counts, keyword results, top words) as JSON, or a keyword,count,density table as CSV — both copied to the clipboard.
One primary term, matched as an exact phrase. Optional — without it you still get statistics and tables.
Comma-separated additional terms, each analyzed like the focus keyword.
Switches common function words back into the repeated-word and phrase tables. Off by default.
Analysis runs on click rather than live; Clear resets the text, keywords, and results.
Percentages are unstable on small samples — three mentions in fifty words is 6%. The tool warns below 25 words; judge short snippets by reading them, not by the percentage.
Matching is exact after normalization: word order matters and word forms are not stemmed, so brewing cold does not match cold brew, and neither does brews. Check for a typo or try the singular form.
Stop words are filtered out by default, and single-character tokens are skipped. Flip the include-stop-words switch to see the unfiltered pattern.
Matching is exact-phrase after normalization — there is no stemming or synonym detection, so brew, brews, and brewing are three different words to the counter. The status bands are general editorial guidance, not thresholds any search engine publishes; no density percentage is a ranking target. The checker analyzes pasted text only — it does not fetch URLs, so run it on the copy itself rather than a link. Very short texts produce unstable percentages, which the built-in warning flags below 25 words.
Your draft is analyzed entirely in your browser when you click Analyze — the text, keywords, and results are never uploaded or stored by the tool, which makes it safe to review unpublished or client content.
Treat the density bands as review guidance, not targets — read any flagged section aloud before editing.
Analyze complete drafts rather than fragments; percentages are unstable on short samples.
Use the phrase tables to find repetitive wording patterns, not just keyword counts.
Cover the topic with related terms and synonyms where they help the reader, not to hit a number.
Re-run the analysis after editing — results update only when you click Analyze.
Run the analysis after writing, not during. Drafting against a live percentage produces copy shaped by a number; drafting for the reader and then checking produces copy that is merely verified. When a keyword lands in the high or overused band, the fix is usually structural — a heading that repeats the title, a boilerplate sentence pasted into several sections — rather than word-by-word substitution.
Read the tables as a mirror of your habits. The two- and three-word phrase tables catch the constructions you reach for unconsciously; pruning those improves the writing whether or not any search engine is watching.
And keep the goal in order: search engines reward pages that treat a topic thoroughly and readably. Density checking helps you notice when wording gets in the way of that — it is quality control for prose, not an optimization lever. Pair it with the Word Counter for length targets and the SERP Snippet Preview for how the page will present in results.
Tokenization lowercases, applies Unicode NFKD normalization to strip accents, removes apostrophes, converts hyphens, underscores, and slashes to spaces, and drops remaining punctuation (Unicode-aware, so non-Latin letters survive). Phrase counting is a sliding-window exact match over the token array. N-gram tables use two- and three-token windows, keeping entries with count above one; with stop words excluded, a phrase must contain at least half meaningful words. Reading time is total words divided by 220, rounded up with a one-minute floor. Analysis is memoized on the submitted snapshot and runs wholly client-side.
| Density | Status | Suggested reading |
|---|---|---|
| 0% | Not found | The keyword never appears |
| below 0.5% | Low | Present but rare — add only where it helps readers |
| 0.5–2.5% | Natural | Typical range for most articles |
| 2.5–4% | High | Review manually — may still be fine |
| above 4% | Possibly overused | Likely reads as repetitive |
The bands are editorial guidance for review, not published ranking thresholds — always confirm by reading the flagged text.
Content writers and editors doing a pre-publish pass, SEO teams reviewing keyword coverage across drafts, freelancers documenting keyword usage for clients via the exports, and anyone who suspects a piece of copy repeats itself but wants numbers instead of a hunch. It is a review instrument — most useful to people who will read the flagged passages and judge them, rather than chase a percentage.
No — search engines publish no such number, and modern ranking is not driven by keyword frequency. The bands here (natural around 0.5–2.5%) describe what typically reads well, which is the thing worth optimizing.
Yes — the focus and secondary fields accept phrases, matched as exact consecutive words after normalization. "keyword density checker" is counted only where all three words appear in order.
A built-in stop-word list keeps common function words out so the tables surface meaningful terms. Flip the include-stop-words switch when you want the complete pattern.
No — there is no stemming, so brew and brews are separate entries. Check important variants as secondary keywords if you need each measured.
The analysis runs when you click Analyze content, not live — deliberate, so results stay stable while you edit. Click the button again after changes.
Copy JSON produces a structured summary — the statistics, focus and secondary keyword results, and top ten words with densities. Copy CSV produces a keyword,count,density table of the top repeated words. Both go to your clipboard.
Yes — the analysis runs locally in your browser and the text is never transmitted or saved. Closing the page discards everything.