Is 3% a Google limit?

No. It is an editorial flag, and we only apply it when there are at least 100 words.

No. Counts are exact tokens. starter, starters, and starter's are separate rows.

What is the word count?

Every token in the pasted or extracted text, including a, the, and numbers. Phrase tables hide those; the percentages still use the full total.

Why are Wikipedia references missing?

Reference lists, citation templates, and number-like tokens are stripped by default so the tables show article language.

When is JavaScript blamed?

Only if the HTML is large and almost empty of words. Plenty of HTML text with a thin extract is a selector miss, not JavaScript.

Does this show keyword volume?

No. We count phrases on the page you gave us. We do not invent monthly searches.

Keyword Density

Loading tool…

About Keyword Density

Choose Page URL or Paste text, optionally add a focus keyword, and click Count phrases. Exclude citations and numbers is on by default so reference lists, DOIs, and years do not dominate the tables. You get single words, two-word phrases, and three-word phrases ranked by count, plus density as a percent of every token in the text.

This is not search-volume data. Tokens are exact: starter, starters, and starter's are separate rows. Two-word and three-word phrases must appear more than once to show. Over 3% is flagged only when there are at least 100 words. Function words such as the and and stay in the word count but are hidden from the phrase tables.

How to use Keyword Density

1Choose Page URL or Paste text. Optional: a focus keyword to measure that phrase exactly.
2Leave Exclude citations and numbers on. Turn it off if the copy is numeric (versions, years, standards).
3Click Count phrases. Single words, two-word, and three-word tables, ranked by count. Density is count divided by every word in the text.
4Rewrite stuffed sections. Then Count phrases again.

How phrases are counted

A URL uses main or article text after stripping nav, footer, Wikipedia infoboxes, and reference sections. If the HTML is large but almost empty of words, we warn that the page may be JavaScript-rendered. If the HTML already has plenty of text and we still extract little, that is a selector miss, not JavaScript. Paste mode counts the block as given and does not run those warnings.

Focus keyword is measured on the full token list, including a numeric phrase if you searched for one. Paste is capped at 100,000 characters. Tables preview the top 25 unigrams and top 15 longer phrases, with Show all when you need the rest.

Limits and privacy

URL mode fetches a public page. Localhost and private addresses are blocked. Paste is sent for that count only. Five counts per hour per IP on this tool, including paste. A short cache may reuse a recent result. There is no account. The old /seo/keyword-density-checker URL 301s here.

FAQs

Is 3% a Google limit?: No. It is an editorial flag, and we only apply it when there are at least 100 words.
Do you stem plurals?: No. Counts are exact tokens. starter, starters, and starter's are separate rows.
What is the word count?: Every token in the pasted or extracted text, including a, the, and numbers. Phrase tables hide those; the percentages still use the full total.
Why are Wikipedia references missing?: Reference lists, citation templates, and number-like tokens are stripped by default so the tables show article language.
When is JavaScript blamed?: Only if the HTML is large and almost empty of words. Plenty of HTML text with a thin extract is a selector miss, not JavaScript.
Does this show keyword volume?: No. We count phrases on the page you gave us. We do not invent monthly searches.