Keyword Density Checker

Keyword density is the percentage of times a word or phrase appears in a piece of content relative to the total word count. The formula is: (keyword occurrences / total words) × 100. A 1,000-word article that uses "running shoes" 10 times has a density of 1% for that phrase. This tool calculates density per 1-, 2-, 3-, 4-, and 5-word phrase across your content, with stopword filtering and apostrophe normalization.

No. Google has stated publicly multiple times that keyword density is not a ranking factor. The signal stopped mattering in any meaningful way more than a decade ago, when search engines moved from term-frequency matching to semantic understanding. So why does this tool exist? Because keyword density is still a useful diagnostic: extreme outliers (way too high or way too low) often correlate with real on-page issues like thin content, keyword stuffing, or topic drift. We surface the density numbers but pair them with what actually matters today: real monthly search volume for every phrase, and a coverage gap analysis against the top-ranking pages on Google.

There is no objectively correct keyword density. The conventional advice of "1-2%" comes from early-2000s SEO and has no empirical basis in modern Google. A more useful guideline: write naturally, cover your topic comprehensively, and worry about density only if a phrase is appearing so often it feels forced (above ~3-4% for any single keyword is usually a red flag for keyword stuffing). This tool flags the over-emphasized phrases automatically so you can spot them at a glance.

When you toggle Compare with top Google results, we fetch the top-5 organic results from Google's SERP for your focus keyword (via a paid third-party SERP API), then proxy-fetch each result's actual page content. We then extract all 1-, 2-, and 3-word phrases from those competitor pages and diff them against your content. The output is three panels: Missing topics (phrases your competitors cover that you do not), Over-emphasized (phrases you cover more than your competitors do), and AI Overview phrases (phrases mentioned in Google's AI Overview for that query, if one exists). All of these are sorted by real monthly search volume so you can prioritize the gaps that actually drive traffic.

• Check a live URL: we fetch your published page through our proxy pool, parse the HTML, and analyze the body content. Best for auditing what's actually live.
• Paste content: paste raw text, markdown, or HTML directly. Useful when you want to analyze a draft before publishing, or content that lives behind a paywall or login.
• Live editor: a rich-text editor that scores your content in real time as you type. No server round-trip per keystroke; the tokenizer runs entirely in your browser. Best for actively writing and tuning a piece.

If a phrase is over-emphasized (red badge), replace some occurrences with synonyms, pronouns, or related variations. If a phrase is missing entirely (showing in the Missing topics panel against your competitors), look for natural places in your existing structure to weave it in: subheadings, intro/outro, body paragraphs. Avoid stuffing the same phrase repeatedly; semantic variation is what modern search engines reward.

Words like the, a, and, of, to, is, in are grammatically essential but carry no topical meaning. They appear in every English text with roughly the same frequency regardless of topic. Counting them would dominate the density table with noise. We filter ~150 common English stopwords from 1-gram counts by default; toggle them back on if you want the raw frequencies. For 2-grams and longer phrases we apply a strict filter that drops any phrase where ANY token is a stopword (so "the best" and "for you" are both filtered).

The tokenizer and density math work for any language whose text uses whitespace as a word boundary (most Latin-script, Cyrillic, Greek, and many others). The stopword list shipped is English-only, so non-English analyses will surface many filler words in the 1-gram table that you would normally filter; use the SERP comparison feature anyway, which still works because it compares phrase frequencies regardless of stopword status. For languages without spaces (CJK), the tokenizer treats whole sentences as single tokens, which limits usefulness.