A 1,000-word post mentioning its keyword 47 times (4.7%) reads like a robot wrote it — because the writer optimized for a number instead of a reader. Google's stance is explicit: there is no ideal keyword density, and stuffing violates spam policies. So what is density good for? Diagnostics. This guide shows how to check keyword density, read the readout, and fix stuffing without chasing mythical percentages.
Part of the word counter guide. Analyze any draft in the density tool (1–4 word phrases, stop-word filtering, local only); count in the word counter.
How to check density (the right way)
- Paste the rendered text — not HTML source (tags pollute counts). Strip navigation, footers and comments first; analyze body copy only.
- Read 1–4 word phrases: single words show topics; 2–4 word phrases show real targeting (“word counter free” vs scattered “word”, “counter”, “free”). Stop-word filtering removes noise.
- Interpret, don't target: 12 mentions in 1,000 words = 1.2% — pure math. Ask whether the phrase appears where readers need it (title, intro, headings, conclusion), not whether the number hits a myth.
Fixing stuffing (before/after patterns)
| Stuffed pattern | Natural rewrite |
|---|---|
| “Our free word counter tool counts words free online with our word counter…” | “Paste your draft below — counts appear instantly, no signup.” |
| Keyword in every H2 verbatim | Synonyms + pronouns in half the headings |
| City/service list paragraphs | One location page per area, linked — not listed |
- Synonym test: if replacing the keyword with “it” in three spots improves flow, those spots were stuffed.
- Intent test: does each mention answer something? Mentions earning their place stay; decorative ones go.
- Topical coverage beats repetition: related terms (readability, character count, reading time) signal depth better than the 15th exact repeat.
Why the “ideal %” myth refuses to die
Early-2000s SEO tools needed a dial to display, so vendors invented 2–3% targets — then everyone cited everyone else until myth became doctrine. Google's Cutts (2011) and Mueller (2022) both refused magic numbers; modern systems rank topical authority and intent satisfaction, where repetition has diminishing returns after natural mentions. Tool defaults showing % are fine as readouts; articles prescribing % as rules are not. Pair density checks with readability per Flesch guide — one paste, both signals, zero myths.
Entity coverage: what replaced density chasing
Modern ranking rewards covering an entity's full attribute set — for “word counter,” that's character counts, reading time, readability, density, limits — not repeating one phrase. Audit method: list the 8–12 sub-questions PAA and autosuggest surface, verify each gets a genuine paragraph (not a keyword mention), and let density fall where honest coverage puts it (typically 0.5–2% on primaries without trying). Pages ranking today read topically complete because they answer completely; density is the exhaust, never the fuel. When competitors outrank you, compare their sub-question coverage against yours before touching a single keyword — the gap is almost always coverage, not percentage points.
General writing guidance only. Test with real readers; formulas ignore tone and expertise. SEO outcomes vary — no count, score or density promises rankings.