Online Tool Store
Readability Checker
Score text with Flesch, Flesch-Kincaid, Gunning Fog, SMOG, Coleman-Liau and ARI, plus hard sentences, passive voice, adverbs and wordy phrases highlighted live.
🔒 This tool runs entirely in your browser. Your files are never uploaded to a server.
Example shown — two paragraphs saying much the same thing, at very different reading levels.
Flesch Reading Ease
—
Every formula, and what each one needs
| Formula | Score | Reads as | Uses syllables |
|---|
Show the exact formulas
- Flesch Reading Ease
- = 206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words)
- Flesch–Kincaid Grade
- = 0.39 × (words ÷ sentences) + 11.8 × (syllables ÷ words) − 15.59
- Gunning Fog
- = 0.4 × [(words ÷ sentences) + 100 × (3+ syllable words ÷ words)]
- SMOG
- = 1.043 × √(3+ syllable words × 30 ÷ sentences) + 3.1291
- Coleman–Liau
- = 0.0588 × (letters per 100 words) − 0.296 × (sentences per 100 words) − 15.8
- Automated Readability Index
- = 4.71 × (letters ÷ words) + 0.5 × (words ÷ sentences) − 21.43
Exactly what runs above on your text — nothing rounded or simplified for display.
The sentences doing the damage
Longest first. Sentence length is what these formulas react to most.
Full text, highlighted
Click a label to turn that highlight on or off — heuristics, like the syllable count, not grammar-perfect.
What these formulas actually measure. Word length and sentence length, and nothing else. They cannot see whether your argument follows, whether a term is defined before it is used, or whether the text is true — a page of short words in short sentences can score beautifully and be incoherent. They were also calibrated on English school and military texts decades ago, so treat the grade levels as a rough comparison between drafts rather than a measurement of your reader.
How to use it
- Paste your draft. Scores update as you type.
- Read all six formulas rather than one - they disagree, and the disagreement is informative.
- Look at the longest sentences listed underneath; that is usually where the score is made.
- Scroll to "Full text, highlighted" to see hard and very-hard sentences, passive voice, adverbs and wordy phrases marked directly in your own paragraphs, not just counted. Hover a wordy phrase for a simpler alternative.
- Click a legend label to switch that highlight on or off if one colour is getting in the way of another.
- Edit, and watch whether the number moves.
Two ways to say the same thing
The sample loaded above is one idea written twice, with exactly seventeen words each time. The scores are not close:
dense plain words 17 17 sentences 1 3 syllables 56 20 Flesch Reading Ease -89.1 101.6 Flesch–Kincaid 29.9 0.5 Gunning Fog 32.7 4.6
Same word count, 190 points of reading ease apart. Everything the formulas react to is in the other two rows: three sentences instead of one, and twenty syllables instead of fifty-six. That is the whole mechanism - shorter sentences, shorter words. Counting syllables is a guess, so two formulas avoid it entirely. Flesch, Flesch–Kincaid, Gunning Fog and SMOG all need a syllable count, and English spelling does not reliably tell you one. The heuristic here counts vowel groups and then corrects for the common traps: a silent final e (phone counts as 1 syllable, but table counts as 2), -ed endings (jumped counts as 1, but wanted counts as 2), -tion counting as one syllable (information is 4, not 5), and ea counting as one (bread and please are each 1). Measured against a hand-checked list of sixty-two words it gets sixty right - 96.8%. The two it misses, "international" and "unfortunately", both have a silent e buried in the middle and come out one syllable too high. That is honest performance for a spelling-based heuristic, and it is why the tool publishes the number rather than implying the count is exact. Coleman–Liau and the Automated Readability Index sidestep the problem entirely by counting letters instead of syllables. They are shown alongside as a cross-check: when the syllable-based and letter-based grades land more than two apart, something about the text is unusual and neither family should be taken too literally. Where sentences begin and end is also a guess. Every one of these formulas divides by the number of sentences, so miscounting them moves every score at once. Splitting on full stops is the obvious approach and it is wrong the moment an abbreviation appears - "Dr. Smith arrived." becomes two sentences, halving the words-per-sentence figure for that stretch of text. The splitter here keeps a list of the usual offenders, and treats a single initial like the J in "J. R. Smith" as never ending a sentence. Hard and very-hard sentences get their own colour. Any sentence over 20 words is shaded as "hard"; over 30 words, "very hard". That is our own threshold, chosen because working memory starts straining around the 20-word mark in most style guidance - it is not a reproduction of any other tool's undisclosed scoring, just a plain word count applied consistently. Each highlight has its own checkbox above the text, so you can isolate one colour at a time instead of reading five overlapping tints at once. Passive voice, adverbs and wordy phrases are flagged by pattern, not by grammar. The highlighted view marks a form of "to be" followed by a past participle - "was written", "is being reviewed" - as passive voice, and a word ending in "-ly" as an adverb, skipping a short list of common exceptions such as "family", "apply" and "only". Neither check is a real grammar parser: "was ready" is correctly left alone because "ready" is not a participle, but the underlying rule is still pattern matching, not full sentence understanding. Passive voice is also not an error - it is often the clearer choice when the actor is unknown or unimportant - so treat both counts as something to glance at, not a score to minimise. A third pass checks for common wordy phrases - "due to the fact that", "at this point in time", "in order to" - against a small plain-English list and marks them purple; hover one to see the shorter alternative. What a good score does not tell you: these formulas count word length and sentence length. That is all. They cannot see whether the second paragraph follows from the first, whether you used a term before defining it, whether the examples help, or whether any of it is correct. Text that is deliberately shortened to game a score usually reads worse, not better - chopping a clear thirty-word sentence into three stubs improves the number and damages the writing. The scales are also old and narrow. Flesch–Kincaid was developed for the US Navy in 1975, SMOG for health materials in 1969, and all of them were fitted to English prose of that era. Use them to compare two versions of your own text, which is the comparison they are genuinely good at, rather than as a measurement of your reader.
FAQ
What score should I aim for?
For general audiences, a Flesch Reading Ease around 60 or above and a grade level around 8 to 10. Technical documentation for specialists can sit lower without being a problem. The more useful comparison is between two drafts of your own text rather than against an absolute target.
Why do the formulas disagree with each other?
Because they weight the same two ingredients differently and were calibrated on different material - school textbooks for Flesch–Kincaid, technical writing for Gunning Fog, health leaflets for SMOG. When the syllable-based and letter-based families land more than two grades apart, that is a sign the text has an unusual shape, and the tool says so.
How accurate is the syllable counting?
Good, not perfect. Counting syllables from spelling is a heuristic, and the one here scores 60 out of 62 on a hand-checked word list. The two it misses are words with a silent e in the middle, like "unfortunately", which it counts as six rather than five. Long words also carry the most weight in these formulas, so errors there matter more than they should.
Can I trust a good score?
Only as far as it goes. These formulas measure word length and sentence length. They cannot tell whether your argument follows, whether you defined a term before using it, or whether anything you wrote is true - a page of short words in short sentences can score beautifully and be nonsense.
Does the Flesch score really run from 0 to 100?
That is the conventional range, not a limit. The formula is unbounded in both directions: the dense sample on this page scores −89.1, and the plain one scores 101.6. Anything outside 0–100 simply means the text is far past the material the scale was built on.
Is my text sent anywhere?
No. Everything is counted in the page as you type. Nothing is uploaded, stored or logged, which matters if you are checking a draft that is not public yet.
How reliable is the passive-voice and adverb highlighting?
It is a pattern match, not a grammar parser: a form of "to be" followed by a past participle is marked passive, and a word ending in "-ly" is marked an adverb, skipping a short exceptions list for common false positives like "family" and "only". It gets ordinary sentences right, but it can miss unusual phrasing a real parser would catch. Passive voice is also not a mistake to eliminate - sometimes it is the clearer choice - so use the count as a prompt to look, not a target to hit zero.
What counts as a "hard" or "very hard" sentence?
Sentence length, and only that: over 20 words is shaded hard, over 30 very hard. It is a plain word count, chosen because most plain-language guidance flags roughly that range as where a reader starts losing track of a sentence - it is not the same formula any other readability tool uses internally, since none of them publish theirs.
Can I turn off individual highlights?
Yes. Each colour - hard sentence, very hard sentence, passive voice, adverb, wordy phrase - has its own checkbox above the highlighted text, so you can look at one kind of issue at a time instead of five colours competing for attention.
What is a "wordy phrase," and where does the suggestion come from?
A short list of common padded phrases with a plain-English alternative - "due to the fact that" instead of "because," "at this point in time" instead of "now," "in order to" instead of "to." Hover a highlighted phrase to see the suggestion. It is a fixed list, not a language model, so it only catches phrases already on it.
What is the difference between Flesch Reading Ease and Flesch-Kincaid Grade Level?
Same two ingredients - words per sentence and syllables per word - put through different arithmetic. Flesch Reading Ease outputs a 0–100 score where higher means easier. Flesch-Kincaid Grade Level rescales the same inputs to a US school-grade number, so an 8th-grade result and a Reading Ease score around 60–70 describe roughly the same text. This tool shows both because people search for each by name and the mapping between them is not exact.
Does a readability score affect SEO rankings?
Not directly - there is no confirmed Google ranking factor tied to any of these formulas. What correlates is indirect: clearer writing keeps readers on the page longer and gets shared and linked to more, which are things search engines do reward. Use the score to make a draft easier to read, not as a number to game for its own sake.
How we compare
| Feature | Online Tool Store | Hemingway App | WebFX Read-Able Test | charactercalculator.com |
|---|---|---|---|---|
| Six named formulas at once | ✓ | One blended score | ✓ | Two formulas |
| States its own syllable-counting accuracy | ✓ | ✗ | ✗ | ✗ |
| Cross-checks syllable-based scores against syllable-free ones | ✓ | ✗ | ✗ | ✗ |
| Highlights hard/very-hard sentences, passive voice, adverbs and wordy phrases live, in your own text | ✓ | Highlighted, not independently toggleable | ✓ | ✗ |
| Each highlight colour can be switched on or off independently | ✓ | ✗ | ✓ | No highlighting to toggle |
| Prints the exact formula behind every score | ✓ | Single blended score, formula not shown | Not shown | ✓ |
| Nothing typed leaves the page | ✓ | Not confirmed | Not confirmed | ✓ |
Six named scores with their exact formulas printed, hard/very-hard sentences, passive voice, adverbs and wordy phrases marked directly in your own paragraphs with each colour independently toggleable, and your draft never leaving the page. It will not rewrite anything for you and it cannot tell you whether the writing is good - for that you need a reader.