Keyword Density in 2026: What the Number Is Actually Good For
Why keyword density stopped being a ranking factor, which percentage range still makes sense, and how to read one-, two-, and three-word phrase lists.
Someone will tell you the magic number is 2%. Someone else swears by 1.5%, and there’s a forum thread from 2009 that still insists on 3 to 5%. They’re all guessing.
Google has never published a target. John Mueller has been asked about keyword density on Search Central calls more times than any person should have to answer the same question, and the answer hasn’t moved: it isn’t something the ranking systems look at. Panda, back in 2011, was pointed squarely at the pages that treated it as a lever.
So why does this site have a Keyword Density Checker at all?
Because it’s a decent smoke alarm
Density is a bad target and a useful symptom. Nobody should write to hit 1.8%. But when you paste a 1,600-word article in and the phrase you’re supposedly ranking for shows up twice, at 0.13%, that’s worth knowing before you publish. The page isn’t about the thing you think it’s about. Not to a reader, and not to a language model chewing through it either.
The opposite case is louder. Local service pages are the classic offenders: 400 words of copy with the money phrase wedged in eighteen times, which comes out at 4.5%. You can feel that while reading. The number is what makes it arguable in a meeting.
The math is less obvious than it looks
For a single word, density is just count divided by total words — and that denominator is the word count, counted exactly the way this checker counts it, so the two tools always agree on the total. Fine. For a phrase, most tools quietly keep the same denominator, which makes three-word phrases look artificially rare compared to single words.
This checker divides by the number of positions a phrase of that length can actually occupy: total words minus one for a two-word phrase, minus two for a three-word phrase. In a 1,000-word article the difference is under half a percent, so it’s not going to change your life. It does mean the three tabs are comparable to each other, which is the point of having three tabs.
Read the two-word list first
The single-word table is mostly a vocabulary report. Filter the stop words out and you get your topic nouns, which you already knew about. If single words are all you need — a full ranked table of every word with counts and percentages, plus a CSV export — the Word Frequency Counter is the page built for that.
Phrases are where you learn something. Two- and three-word entries show what the page actually claims, and they surface the tics you can’t see in your own draft. Every writer has them. Mine is “in practice” — invisible to me on the third read, sitting near the top of the two-word table the moment a machine counts it.
Where the number lies to you
A few things it can’t know:
- Inflected languages split the count. In Russian, “плотность”, “плотности” and “плотностью” register as three unrelated words, so a genuinely well-covered topic can look thin.
- Synonyms and paraphrases count for nothing here, but they count for a lot in how search engines read a page.
- Navigation and footers, if you paste a whole page, will bury the real content under sitewide boilerplate. Copy the article body only.
Set the minimum occurrences to 2 or 3, keep the stop-word filter on, and treat anything above 2.5% as a line to reread rather than a rule you broke. Paste a draft into the Keyword Density Checker and see what you’ve been repeating without noticing.