Density is not a ranking factor. Stuffing still tanks rankings.
Google has publicly said since 2011 that keyword density is not a direct ranking signal. Every serious SEO repeats that line. What they skip: extreme repetition still trips Google's spam-detection classifiers, keyword stuffing is a named violation in Google's spam policies, and the March 2024 core update specifically downranked pages with unnatural keyword repetition. So density matters, just not the way keyword-stuffing tools from 2012 framed it. The right mental model is over-optimization detection: use density as a red-flag check, not a target to hit.
There is no single safe percentage, because natural repetition depends on page type. On a product description, repeating the product name 4 percent of the time is normal; the page is literally about that product. On a blog post, 4 percent on a single term reads forced to any human reader and signals over-optimization to Google. Landing pages sit in the middle. The checker above uses 0.5 to 2.5 percent as its good range, the conservative bracket that fits blog and editorial content, so treat a high flag on a product page as a prompt to re-read the copy rather than an automatic failure.
Bigram and trigram analysis matters more than single-word density for most commercial pages. Your target is almost never a single word; it is a phrase like trail running shoes or custom shopify theme development. The raw frequency of shoes on the page might be fine while the exact three-word phrase repeats far too often, and that exact-phrase repetition is the real over-optimization signal. The checker shows all three n-gram tables, never counts phrases across sentence boundaries, and tracks each comma-separated target keyword with its own density, status chip and first-paragraph check.
Density and readability interact. Pages built from long sentences and polysyllabic jargon tend to get stuffed during rewriting as editors push keyword presence, while simple, fluid prose absorbs keyword variation naturally. After tightening density here, run the same text through our readability checker to score it on Flesch reading ease and Flesch-Kincaid grade level in one pass.
Tools in the same cluster: the meta tag generator to write the title and description once density is under control, the headline analyzer to pressure-test the H1, and the schema markup generator to ship structured data.
Frequently asked questions
What keyword density is too high?
There is no single safe number, which is why most tools refuse to give one. A workable calibration by content type: on a blog post or editorial article, anything above 2.5 percent on a single target term deserves a rewrite; on a landing or service page, 3 percent; on a product description, 4 percent, since catalog pages naturally repeat the product name more. These are over-optimization warning thresholds, not ranking formulas. Google uses thousands of signals and has publicly said density is not a direct ranking factor, but extreme repetition still triggers spam-detection heuristics. The checker above flags any target keyword over 2.5 percent, the strictest of those brackets, so apply your own judgment on product pages.
Are bigrams and trigrams more important than single-word density?
Usually yes. Most commercial keywords are multi-word phrases (trail running shoes, custom shopify theme development, ecommerce growth strategy), so single-word density often mis-reads natural vocabulary as keyword stuffing or the other way around. The 2-word and 3-word tables surface real phrase-level repetition. If your target is trail running shoes and the 3-word table shows that exact phrase appearing 18 times in 1,200 words, that is the number to check against the threshold, not the raw count of shoes. The tool shows all three n-gram levels so you can match the analysis to your target phrase length.
Should I exclude stop words?
Yes by default. Stop words (the, a, an, is, of, in, to, and similar grammatical words) are the highest-frequency tokens in any English text and flood the density tables without telling you anything useful about topic focus. The checker filters a standard English stop list by default. Turn it off when debugging unusual text (legal documents, poetry, transcripts) where stop-word distribution carries meaning, or when analyzing non-English content: the tokenizer handles non-English words, but the stop list itself is English-only, so for German, Spanish, or other languages disable filtering and interpret manually.
What is a good Flesch reading ease score?
Flesch reading ease runs 0 to 100, higher is easier. Target brackets by content type: marketing blog 60 to 70 (8th to 10th grade reading level, which is where most US adults comfortably read); technical documentation 50 to 60 (10th to 12th grade); legal or medical 30 to 50. Below 30 the text is college-level and harder to skim. Above 80 reads simple and often choppy. The formula penalizes long sentences and polysyllabic words, so the fastest way to raise a low score is to break long sentences and replace 4-syllable jargon with 2-syllable synonyms. This page does not compute the score; paste the same text into our readability checker to get Flesch ease and grade level.
Does this tool save my text?
No. Every word you paste lives in memory for this browser tab only. Nothing is transmitted to a server, stored in a database, or synced. The Copy report button puts a plain-text summary on your clipboard, and Download CSV writes the n-gram and target tables to a local file; those are the only two output paths. The page uses localStorage only for your theme and cookie-consent choices, never for the text you analyze, and you can clear both by wiping browser data.