How To Analyze Keyword Density

The basic calculation

Keyword density is the percentage of a page’s total words that match a target term. It was once the single most discussed metric in SEO, the axis on which whole chapters of Google’s 2003 ranking algorithm turned. A decade of deliberate stuffing penalties, the Panda update, BERT, and the shift to semantic search has stripped density of most of its direct ranking power. But it remains useful as a diagnostic: it tells you whether a page is actually about what you think it is about, whether you are under-serving the primary intent, or whether the language has drifted into padding. This guide covers the basic calculation, historical versus modern SEO use, penalties for overuse, stemming and variation handling, what “natural frequency” means, and where LSI keywords actually fit.

Historical SEO: what density meant

Keyword density is (occurrences of the keyword / total words) × 100. A 1,000-word article that mentions “email marketing” 15 times has an exact-match density of 1.5 percent. For a single-word keyword the counting is simple. For multi-word phrases the numerator stays the count of phrase occurrences, and the denominator stays total words, which dilutes density because one occurrence of a two-word phrase still counts as one in the numerator but consumes two slots in the denominator. Most density tools report the phrase count over the word count without adjusting, which is the convention.

Modern SEO: what density means now

In the first decade of search, keyword density was a strong ranking signal. Pages with densities in the 2-5 percent range for a target term outranked pages that mentioned the term only in passing. This led to a decade of over-optimization: thin pages stuffed with target keywords, doorway pages built to rank for single terms, and footer-text gardens designed purely to hit a density threshold. The early search engines treated high density as a positive signal because the alternative was worse—ignoring the term entirely.

Stuffing penalties

Google’s 2011 Panda update, 2013 Hummingbird rewrite, and 2019 BERT and 2022 MUM-style language models have all moved ranking away from word-frequency matching toward semantic understanding. BERT can tell that a page about “running shoes” and a page about “athletic footwear for runners” are about the same topic even without identical keywords. Density is no longer the direct lever it once was. But it is still a useful diagnostic: if your page about “email marketing” never uses the phrase, or uses it only in the title, the page may be less focused than you think.

Natural frequency

Google’s spam policies explicitly list keyword stuffing as a violation. The detection is not tied to a precise density threshold—it looks for patterns that suggest mechanical insertion rather than natural writing. Repeating the keyword in every sentence, inserting it into irrelevant places, or filling footer text and alt attributes with variants all trigger flags. A page with 5 percent density written naturally can rank fine; a page with 2 percent density that reads like robot-generated padding can get suppressed. The modern rule: write for readers, let density follow.

Stemming and partial matches

Natural frequency is the density you get when a knowledgeable writer addresses the topic without thinking about density at all. It varies by topic. Technical content about a specific product mentions the product name often—2 to 3 percent is common and normal. A broader article about a category mentions the category term less often because the writer uses pronouns, synonyms, and partial references. If your target density for a term is wildly above or below what a human writer would produce naturally on that topic, the density is signaling a problem with the writing.

LSI and semantic terms

A density tool that counts only exact matches under-reports the real prominence of a topic. “Email marketing”, “email marketer”, and “marketing emails” all signal the same topic to a modern search engine. Stemming collapses inflected forms (marketing, marketer, markets) to a common root. Lemmatization is stricter and maps word forms to dictionary headwords. Most density tools offer exact-match by default and stemmed-match as a toggle. For SEO analysis, stemmed counts are usually more honest.

Competitor density analysis

Before optimizing your density, measure what top-ranking competitors use. Pull the top five organic results for your target query, strip navigation and boilerplate, and compute density for each. The median tells you what density Google considers appropriate for that query. If your page sits far outside that range (either much higher or much lower), the content is probably an outlier in either information density or relevance. Match the competitive baseline before trying to exceed it.

Density by document zone

Word position matters more than raw density. A keyword in the title, H1, first paragraph, URL, and first image alt attribute signals topic more strongly than the same keyword repeated ten times in a sidebar. Modern density analysis should weight different zones differently, or at least report density per zone (title, H1, intro, body, footer). An unfocused page with the right density in the wrong places underperforms a focused page with lower density where it counts.

Common mistakes

Run the numbers