What is the right keyword density?

There is no right number, and anyone naming a specific percentage as the norm is selling you a twenty-year-old myth. A working reference point is 0.5% to 2.5%. But that is a by-product of normal writing, not a target: write first, check second, never the other way round.

Why does the tool miss my phrase when it is clearly in the text?

Most likely the wording differs slightly, or the words are separated by a comma or another word. The search matches an exact run of words inside a single sentence. Try entering the form that actually appears in your text.

Why bother with two- and three-word phrases when there is a single-word list?

Because people search in phrases. Single words show the vocabulary of the text; pairs and triples show which constructions you repeat. That is often where the problem shows: different words, the same turn of phrase in every paragraph.

What does the stop-word switch do?

It removes function words from the lists: prepositions, conjunctions, pronouns, particles. Without it the top places always go to "the", "of" and "in", and the table becomes useless. In pairs and triples only phrases that start or end with a function word are dropped: "rate per hour" stays, because a function word in the middle belongs there.

Can I analyse someone else's text to understand a competitor?

You can, and it is more useful than analysing your own. Take two or three articles from the first page of results, paste them one at a time and look at the three-word phrase table: you will see which subtopics they cover and which are missing from your text entirely. That is far more valuable than a percentage.

The text is short and the percentages are huge. Is that normal?

Yes, and that is exactly why the tool warns about texts under a hundred words. At that length every use weighs more than a percentage point, and comparing those numbers with a long article means nothing.

Keyword density: what a reader notices as repetition

Paste your text and you see which words, two-word phrases and three-word phrases repeat most often, how many times, and how much of the text they take up. Your target phrase is counted separately. This is not a tool for hitting a percentage. It is a way to see your own text through someone else's eyes: when one word takes up five percent of an article, the reader hears it without counting anything.

Article text

Making money writing starts not with registering on a freelance marketplace but with the plain habit of writing every day. A beginner usually looks for a way to make money writing fast and at a high rate, but it works out differently: first you write cheaply and slowly, then faster and for more. Your first clients come from referrals, so keep every finished article in a portfolio.

Once the portfolio holds at least a dozen pieces, you can raise your rate. A newcomer in 2026 starts at around forty dollars for a thousand words, and a year later a confident author charges four times that for the same work. The difference is not typing speed but how precisely the text solves the client problem.

A narrow topic pays better than being a generalist. Write about what you genuinely understand: medicine, finance, law, engineering. A client pays for understanding the subject rather than for word count, and that is where you make money writing instead of merely filling pages.

Another beginner mistake is writing without a plan. An article with no structure eats more time in revisions, and time is the main resource a freelancer has. Sketch an outline of three or four subheadings and the work moves noticeably faster.

Count your rate per hour, not your fee per article. An article that pays two hundred dollars and takes eight hours pays worse than an article that pays eighty dollars and takes one hour. Beginning freelancers look at the number in the contract and almost never look at their own rate per hour, and that is the number showing whether you are moving forward.

A rate per thousand words is the most convenient way to compare offers. When you see the rate per thousand words from two clients at once, the difference in money becomes obvious and volume stops muddying the picture.

So keep a simple log: date, client, volume, time spent, amount. After three months the numbers will show you which topics and which clients are worth it and which ones to drop.

The text is processed in your browser and is never sent anywhere.

Target phrase Separate several phrases with commas. Each one is counted on its own.

Text language The language only decides which stop-word list is used.

Minimum word length Shorter words are left out of the single-word list.

Processing Remove stop words Prepositions, conjunctions, pronouns. Phrases that start or end with a stop word are dropped.

Target phrases

under 0.5% — the phrase is barely there 0.5–2.5% — working range over 4% — looks like keyword stuffing

Word	Times	Density
Paste some text to see the list.

Two-word phrase	Times	Density
Paste some text to see the list.

Three-word phrase	Times	Density
Paste some text to see the list.

Your text never leaves the browser: the counting and the stop-word lists are built into the page itself.

Keyword density is not a ranking factor on its own — Google has said so directly. This tool shows something else: what a reader will notice as repetition.

How density is calculated

The formula is simple: number of uses × number of words in the phrase ÷ total number of words × 100. A three-word phrase used twice is six words of repetition — that is how the reader hears it, which is why multiplying by phrase length is not optional here. The denominator is every word in the text, function words included. Stop words are dropped from the frequency lists but not from the denominator: take the prepositions out of it and the percentages roughly double, and no comparison with any other text works any more. Phrases do not jump across a full stop. "…a high rate. An author looks…" does not produce the word pair "rate author": the text is split into sentences first, and pairs and triples are collected only inside a sentence. Pairs and triples that occur just once are left out of the table — they show nothing.

Which numbers count as normal

Below 0.5% the phrase is barely there. If the article really is about the topic, it is odd that the topic is never named outright. That is not a reason to bolt the phrase on artificially, it is a reason to reread your opening. Between 0.5% and 2.5% is the working range. That is roughly what text looks like when a person who understands the subject writes it in plain language. Over 4% looks like keyword stuffing. Not because a search engine has some threshold, but because at that frequency the repetition becomes audible when you read out loud. Test it that way: read a paragraph aloud and you will hear what the table shows. These edges are a reference point, not a rule. In a short 200-word text one extra use moves the percentage by a full point, and the numbers there are not worth trusting.

Why density is not a ranking factor

Google representatives have said so directly: modern search has no separate counter for how many times the query appears in a text. The algorithms have worked with synonyms, word forms and the topic of the text as a whole for years, not with the arithmetic of repetition. So what does this tool show? Two things. First, whether the topic is present in the text at all, or whether you wrote an article about earning without once using the word "earning". Second, whether you overdid it in the other direction: text written to a formula reads badly, and people leave it unread. Reader behaviour does matter. A special case is text produced by a language model and left unedited. Its density is usually flat and dull: the same construction repeats in every paragraph. The three-word phrase table makes that obvious at a glance.

What the tool does not do

It does not know word forms. "Earning" and "earnings" are different words to it, and a phrase is matched as an exact run of words. That is deliberate: handling morphology would need a dictionary several megabytes in size, and the tool is not allowed to load anything over the network. It does not read a page from a URL and it sends nothing anywhere: it works only with the text you paste yourself. It does not judge quality, style or readability — there are separate tools for that. And it will never tell you to use the phrase twice more: that advice does not exist here, because that is exactly how unreadable text gets written.

Keyword density: what a reader notices as repetition

How density is calculated

Which numbers count as normal

Why density is not a ranking factor

What the tool does not do

Frequently asked questions