What is a good keyword density? The honest answer for modern SEO

Ask a search engine optimiser from 2008 for the secret to ranking and they'll tell you a number: aim for a keyword density of two or three percent and you're golden. Ask anyone who works in SEO today and they'll wince. The metric is real and still worth checking — but the idea that there's a single magic percentage that lifts you up the rankings has been dead for more than a decade. This piece explains what keyword density actually measures, why it stopped being a ranking lever, and how to use it the way it's still genuinely useful: as a guardrail against over-optimisation.

What keyword density actually measures

Keyword density is simple arithmetic. Take how many times a word or phrase appears, divide it by the total number of qualifying words on the page, and multiply by 100. If "remote work" shows up 14 times across a 500-word post, that's roughly a 2.8% density for that phrase. Nothing mysterious — it's a frequency ratio expressed as a percentage.

The subtlety is in what counts toward "total words." A naive count that includes every "the", "a", and "of" produces useless numbers, because function words always dominate. That's why a good analyser filters stopwords — the high-frequency glue words that carry no topical meaning — before it calculates anything. With the filter on, "the" stops topping your list and the words that actually describe your subject rise to the top, where you can judge them.

Single words tell you less than phrases

One-word density is the version everyone learned first, and it's the least informative. A page about espresso machines will naturally have a high density of "espresso" — that tells you almost nothing about whether you're optimising well or stuffing. The interesting signal lives in two- and three-word phrases.

That's because real search queries are phrases. People don't search "espresso"; they search "best espresso machine for beginners" or "espresso machine descaling". When you look at two-word (bigram) and three-word (trigram) density, you see whether the page is actually built around the phrases people type, or whether a single word is just repeated in unrelated contexts. Our keyword density analyzer reports all three lengths — 1, 2, and 3 words — precisely so you can compare them. A page where the target phrase appears naturally several times is in much better shape than one where a lone word is hammered.

One quirk worth knowing: for multi-word phrases, the analyser only drops a phrase if every word in it is a stopword. "How to" survives because it's a meaningful navigational pattern, while pure filler like "of the" is discarded. That keeps useful instructional phrases visible instead of silently filtering them out.

The number that matters is the upper limit, not the target

Here's the mental shift that separates modern practice from the old myth. There is no density you should be aiming for. There is, however, a density you should not exceed.

The analyser flags two thresholds. Anything above 3% for a given phrase is marked elevated — a soft warning to glance at that phrase and make sure it's there because the writing needed it, not because you were padding. Anything above 5% is flagged high, the zone where text starts to read unnaturally and where search engines have long been able to recognise keyword stuffing. That 5% line is the one to respect. Below it, density is simply not something to optimise; above it, you've introduced a real risk.

So the healthy workflow inverts the old advice. Don't write to hit 2.5%. Write naturally for a human reader, then run the check to confirm nothing accidentally spiked into the elevated or high band. If your focus phrase lands somewhere between 1% and 3% on its own, that's a sign you've used it enough to be unambiguous about your topic without beating the reader over the head — but it's a description of healthy writing, not a goal you forced.

Why density stopped being a ranking factor

For years, search ranking really did lean on term frequency, and so an entire cottage industry sprang up to game it. Predictably, the result was unreadable pages where a phrase appeared every other sentence. Search engines responded by shifting to models that weigh topical coverage and semantic relevance — whether a page discusses the surrounding concepts a genuine expert would mention — rather than counting how often the exact phrase recurs.

The practical consequence: you can no longer move your ranking up by nudging density higher, but you can still move it down by pushing density into stuffing territory. Density became a one-directional signal. It's a tripwire for the bad case, not a dial for the good one. That's exactly why a density checker is still worth running — not to chase a score, but to catch the mistake.

Pitfalls that distort the numbers

• Short pages give unstable readings. On a 120-word post, a phrase used three times already reads as 2.5%+, and one extra mention swings it wildly. Density only stabilises on a few hundred words of real content.
• Inflections count separately. "Design", "designs", and "designed" are three different tokens to a density counter, so your true topical coverage is often higher than any single row suggests. Read related forms together.
• Stopword settings change everything. Turn the stopword filter off and your list will be topped by grammar words that mean nothing for SEO. Keep it on unless you have a specific reason not to.
• Density is one lens, not the whole picture. A raw frequency count or an n-gram breakdown answers different questions. Use density specifically when the metric you want to defend is "what percentage of this page is this phrase."

A simple routine that works

1. Write the page for a human first. Don't think about keywords while drafting.
2. Run the finished draft through the density check at the 1-, 2-, and 3-word levels.
3. Confirm your focus phrase appears at all, and that it sits comfortably below the 5% line — ideally without ever tripping the elevated flag.
4. If anything is flagged high, find those sentences and rewrite for variety, not for the score. Synonyms and related concepts do more for modern SEO than repetition ever could.
5. Re-run, and stop when the writing reads naturally and nothing is flagged. That's the finish line — not a target percentage.

Keyword density survives as a useful tool the moment you stop treating it as a goal. Used as a guardrail, it catches the one mistake that still genuinely hurts — stuffing — while leaving you free to write for the reader, which is what actually ranks now. When you want the numbers without doing the arithmetic by hand, the keyword density analyzer breaks down single words and two- and three-word phrases, with the over-optimisation flags built in so the only thing you have to watch for is the warning.

General information, not advice. Search ranking systems are proprietary and change frequently; treat density as a writing guardrail rather than a guaranteed ranking signal, and prioritise content quality and topical relevance over any single metric.