Readability Grader

No. All analysis runs inside your browser using JavaScript. Your text never leaves your device and nothing is logged. You can verify this by disabling your internet connection — the grader keeps working.

It depends on your audience. For general web content, aim for 60–70. Marketing copy should hit 70+. News articles sit naturally at 55–65. Technical documentation at 45–55 is fine because jargon is expected. Scores above 80 usually feel childish to adult readers.

They measure different things. Flesch-Kincaid uses sentence length and syllable count — every 3-syllable word counts equally. Gunning Fog excludes common suffixes like -ed, -es, -ing, so it penalizes technical jargon harder. A 2-grade difference between them is normal. If they disagree by 4+ grades, your writing style is uneven.

Use Flesch Reading Ease as the primary metric. Google has never confirmed readability as a ranking factor, but analyses of page-1 results consistently show an average FRE of 75+. Target 60–70 for most SEO content and 75+ for landing pages.

For text under 100 words, formulas become noisy — one long sentence can swing the score by 20 points. SMOG specifically requires at least 30 sentences for its published accuracy. For short text, use Flesch Reading Ease as a rough guide and trust your ear.

We flag any sentence containing a form of 'to be' (am, is, are, was, were, been, being) followed by a past participle ending in -ed or -en. This catches about 95% of true passive constructions. It may false-positive on adjectival past participles like 'the door is closed' (state, not passive), but it's a useful first-pass filter.

Spanish uses a dedicated formula (Fernández-Huerta) which we compute when you set the language to Spanish. For French, German, Italian, and Dutch, published variants exist (LIX, RIX, Kandel-Moles) but we haven't added them yet. English-only formulas on non-English text produce unreliable scores.

We use a rule-based heuristic that's roughly 85% accurate against a dictionary. It handles common patterns (silent e, vowel pairs, consonant clusters) but may miscount unusual words, proper nouns, and technical terms. For 500+ word samples, individual word errors average out — aggregate scores typically land within 2% of dictionary-based results.

Formulas count every punctuation mark and alphanumeric character. Long code blocks inflate character count without adding natural sentence structure, distorting grade-level scores. Strip code and verbatim quotes before grading your original prose.

It's the simple mean of six grade-level scores (Flesch-Kincaid, Gunning Fog, SMOG, Coleman-Liau, ARI, Linsear Write). Each formula has its own quirks; averaging reduces noise. A spread wider than 4 grades between the highest and lowest suggests your writing style is inconsistent.

No. A score of 90+ reads like children's content — inappropriate for most business and adult contexts. Match the score to the audience. Writing a research abstract at Flesch 80 would undermine credibility. The right question is 'is this the expected level for this audience?' not 'is this higher?'

Hemingway focuses on one grade-level score plus highlighting hard sentences. Grammarly scores readability as part of a broader grammar/style check. Our tool runs six different readability formulas at once, plus detects passive voice, adverb density, and reading time — all client-side, no signup, free.

Most -ly adverbs signal weak verbs. 'She ran quickly' loses to 'She sprinted.' 'He spoke loudly' loses to 'He shouted.' Not every -ly word is wasted — 'only,' 'really,' 'early' aren't adverbs of manner — but high adverb density usually means your verbs are carrying less weight than they should.

Yes — the formulas don't know or care where text came from. AI-written content tends to score in a narrow band (Flesch 55–65, grade 9–11) because LLMs optimize for a middle-of-distribution style. If your AI draft scores outside that range, it's usually because you specifically prompted for it.

We use 238 words per minute for silent reading, based on Brysbaert's 2019 meta-analysis of 190 studies. Speaking time uses 150 words per minute, a common conversational benchmark. Your actual speed varies with familiarity and text complexity — treat these as reference points, not targets.

Completely free, no signup, no usage limits. Since everything runs in your browser, we literally couldn't rate-limit you if we wanted to. The only ads are non-intrusive banner placements — no popups, no paywalls, no email capture.