Method
Readability scores explained
Flesch Reading Ease uses two inputs and can see neither vocabulary nor meaning. What it is good for anyway, and how to read the number honestly.
Readability scores are the most widely used and most widely misunderstood numbers in writing. They are worth having, and they measure much less than people assume.
What the formulas actually do
Flesch Reading Ease, published by Rudolf Flesch in 1948, takes exactly two inputs: average words per sentence, and average syllables per word.
206.835 − 1.015 × (words / sentences) − 84.6 × (syllables / words)
Flesch–Kincaid Grade Level, developed for the US Navy in 1975 to assess technical manuals, uses the same two inputs weighted differently and returns a school grade:
0.39 × (words / sentences) + 11.8 × (syllables / words) − 15.59
That is the whole of it. Two variables. No vocabulary list, no grammar analysis, no notion of whether the sentences are in a sensible order or whether the text says anything true.
Reading the score
| Reading Ease | Level | Typical of |
|---|---|---|
| 90–100 | Very easy, 5th grade | Early readers |
| 80–89 | Easy, 6th grade | Conversational writing |
| 70–79 | Fairly easy, 7th grade | Popular fiction |
| 60–69 | Plain English, 8th–9th grade | Newspapers, most web writing |
| 50–59 | Fairly difficult, 10th–12th grade | Serious magazines |
| 30–49 | Difficult, college | Academic and professional writing |
| 0–29 | Very difficult, graduate | Legal and technical documents |
The score can go above 100 and below 0; both are legal outputs of the formula and both mean the text is outside the range it was calibrated for.
What the formulas cannot see
This is the important part, and it explains almost every case where a good score accompanies bad writing:
- Word difficulty. The formulas count syllables, not familiarity. Cat and ghyll are both one syllable. Because is two syllables and universally known; so is schema.
- Order. Shuffle the sentences of a paragraph into random order and the score does not move at all.
- Cohesion. Whether each sentence connects to the last is most of what makes prose readable, and is entirely invisible here.
- Jargon. A three-syllable technical term scores the same as a three-syllable everyday one.
- Meaning. Nonsense with short sentences and short words scores beautifully.
- Layout. Headings, lists and white space affect real readability enormously and appear in no formula.
Because both formulas are driven by sentence length, they are trivially gameable: split every sentence at its conjunction and the grade level drops by two or three. The text is not easier to read. It is easier to score.
Where the scores are genuinely useful
- As a change detector. The absolute number is weak; the direction is informative. If a section suddenly scores eight grades harder than the rest of a document, something happened in that section — usually a paragraph of stacked subordinate clauses.
- As a compliance floor. Several US states set minimum Reading Ease scores for insurance contracts, and plain-language requirements in health and government communication frequently cite these formulas. If the requirement names a number, the number is what you need.
- As a first-draft alarm. A grade level of 18 on writing aimed at a general audience is a real signal, even though a grade level of 9 does not certify anything.
Other formulas you will meet
- Gunning Fog. Uses sentence length plus the percentage of words with three or more syllables. Punishes technical vocabulary harder than Flesch does.
- SMOG. Built for health materials and generally considered the most reliable of the family for that purpose. Needs at least 30 sentences to be meaningful.
- Coleman–Liau and ARI. Use characters per word instead of syllables per word, which makes them more reproducible, since character counting is exact and syllable counting is not.
- Dale–Chall. Compares against a list of about 3,000 words familiar to fourth-graders. Slower to compute, but the only common formula that knows anything about vocabulary.
How to actually make writing more readable
None of these will show up reliably in a score, and all of them work:
- Put the subject and verb close together and near the front of the sentence.
- Prefer the active voice unless the object is genuinely the topic.
- Give each paragraph one job.
- Turn a sentence containing a list into a list.
- Replace an abstract noun with the verb it came from: make a decision → decide.
- Vary sentence length deliberately. A short sentence after a long one lands.
- Read it aloud. Everywhere you stumble is a sentence to rewrite.
The counter reports Reading Ease and Grade Level alongside average words per sentence, which is usually the more actionable of the three. And because our sentence rules differ from other tools' — we count headings and bullets as sentences — expect our score to differ from theirs on structured documents. The rules are on How we count.
Common questions
What is a good Flesch Reading Ease score?
For general audiences, 60 to 70, which is roughly eighth to ninth grade. Newspapers typically sit around 60 to 65, insurance documents around 40, and academic papers frequently below 30. Some US states require insurance policies to score at least 45.
What is the difference between Reading Ease and Grade Level?
They use the same two inputs — words per sentence and syllables per word — but scale them differently. Reading Ease runs 0 to 100 with higher meaning easier. Grade Level returns a US school grade, so higher means harder.
Can I just shorten my sentences to improve the score?
You can improve the score that way, and you may not improve the writing. The formulas can only see sentence length and syllable count, so chopping every sentence in half raises the number without making the argument any clearer.
Do readability formulas work on all text?
No. They were designed for continuous prose and are unreliable on bulleted documents, tables, poetry, code and dialogue. A slide deck can score as unreadable while being perfectly clear.
Which readability formula should I use?
Flesch–Kincaid Grade Level if you need a single number for a general audience; it is the one most widely required by US institutions. Treat any of them as a rough signal rather than a measurement.
Why does my score differ from another tool's?
Because syllable counting is approximate and sentence boundaries are contested. A tool that does not treat headings as sentences will report much longer average sentences on a structured document, and the score follows directly from that.