What Is Keyword Density? An Outdated Metric, Mostly

Jul 18, 2026 · SEO Glossary

Let's bust the myth first: there is no keyword density percentage that makes a page rank. Not 2%, not 3%, not any number a plugin's traffic-light badge suggests. Google's search team has said versions of this for fifteen years. And yet "what's the ideal keyword density?" remains one of the questions I'm asked most by new clients — usually right after an SEO tool told them their density was "too low."

What is keyword density? It's the percentage of words on a page that are a specific keyword: a 1,000-word page using its target phrase 15 times has a density of 1.5%. The metric is a fossil from early search engines, which ranked pages substantially by counting term occurrences — and it was gamed instantly and mercilessly. Understanding why it died explains most of how modern on-page SEO actually works, which is why the concept is still worth ten minutes of your time.

What is keyword density's actual history?

In the late 1990s and early 2000s, search engines really did lean on term frequency — more mentions signaled more relevance. The result was inevitable: white text on white backgrounds, keyword lists in footers, pages reading like "cheap flights London cheap flights deals cheap flights booking." Search quality suffered, and Google spent the next two decades building systems that made counting obsolete: link analysis to measure authority independent of on-page repetition, then increasingly sophisticated language understanding — synonyms, entities, and models like BERT that parse what a page means rather than which strings it repeats. Somewhere along the way, keyword stuffing flipped from a winning tactic to a spam signal that gets pages demoted. The metric didn't just stop helping; overdoing it started hurting.

Why the myth refuses to die

Three reasons, all instructive. First, tools keep it alive: content optimizers still display density scores and nag you toward targets, because a number — any number — feels more actionable than "write naturally and comprehensively." Second, there's a correlation trap: pages that rank well often mention their topic a reasonable number of times, so people mistake a byproduct of covering a subject for the cause of ranking. Third, it once worked, and SEO folklore has a long half-life — advice from 2009 still circulates through forums and secondhand training with its expiry date scrubbed off.

What actually signals relevance now

The legitimate kernel inside the dead metric: your target term and the concepts around it should, of course, appear on the page. What matters is where and in what company, not how often:

• Placement beats repetition. The keyword in your title tag, your H1, your opening paragraph, and a heading or two does more than twenty extra body mentions combined. Prominent positions are weighted; the four-hundredth word of a paragraph isn't.
• Coverage beats frequency. A page about espresso machines that discusses pressure bars, grind size, portafilters, and descaling demonstrates topical depth no repetition count can fake. Modern systems evaluate whether the page covers what searchers of that query need — which circles back to search intent, the thing density never measured.
• Natural variation is expected. Humans writing about a topic use synonyms, plurals, and related phrasing without trying. Text where an exact phrase repeats with mechanical regularity reads as optimized-for-robots — to robots that were specifically trained to notice.

My working advice to writers: use the keyword where it belongs structurally, then write the best answer to the query and stop counting. If you finish a draft and the target phrase appears zero times, that's worth fixing. If it appears forty times in a thousand words, that's worth fixing in the other direction. Everything between is noise.

The one time density-style thinking still helps

As a stuffing detector, glancing at term frequency retains diagnostic value — especially on older sites. I've inherited plenty of legacy pages from the count-your-keywords era: exact-match phrases jammed into every paragraph, image alt text, and anchor, sometimes at densities north of 5%. Those pages are liabilities under modern spam systems, and a rewrite toward natural language routinely improves their rankings — one of the more satisfying fixes in a content cleanup, because you're deleting words and gaining positions. If you audit an old site, checking the most-trafficked pages for that vintage over-optimization belongs on the list, right alongside the technical basics a site crawl covers — duplicate titles, thin pages, and metadata that reads like a keyword list are usually symptoms of the same era.

What to tell the tool that's nagging you

When a plugin scores your keyword usage orange and demands two more mentions: it's pattern-matching, not measuring ranking probability. Treat such scores as a rough completeness reminder — did you actually mention your topic in the places that matter? — and ignore the percentages entirely. The pages that win are the ones a knowledgeable human would call the most useful result for the query, and no human has ever praised an article for its density. Spend the saved effort on specificity: the examples, numbers, and edge cases competitors are too lazy to include. That's the metric Google can't stop rewarding, because it's the one users actually feel.