Readability score
Flesch reading ease, Flesch–Kincaid grade level and four other indices.
34 words · 3 sentences
Comfortable for a reader at College graduate level
US school grade level
Years of education needed
Favoured for health materials
Uses characters, not syllables
What is driving the score
Comfortable
How it works
Readability formulas all work the same way: they count sentence length and word complexity, then map the result onto a school grade. None of them read your text — they cannot tell whether it is coherent, only whether it is dense.
Flesch reading ease = 206.835 − 1.015 × (words/sentence) − 84.6 × (syllables/word)
Flesch–Kincaid grade = 0.39 × (words/sentence) + 11.8 × (syllables/word) − 15.59 The scale
- 90–100
- Very easy. Children's books.
- 60–70
- Plain English. Most newspapers and good business writing sit here.
- 30–50
- Difficult. Academic and technical prose.
- Below 30
- Very difficult. Legal and insurance documents.
Why several formulas
They disagree, and the disagreement is informative. Coleman–Liau and ARI count characters rather than syllables, so they are immune to syllable-estimation error but blind to the difference between “through” and “thought”. SMOG counts polysyllabic words and is the standard for health materials. When all five agree, the score is trustworthy.
What to actually do with a bad score
Split long sentences. Replace nominalisations with verbs (“make a decision” → “decide”). Cut qualifiers. Choose the shorter word where it means the same thing. Every one of those makes the text genuinely clearer — which is why the score moves.
Do not optimise the number
These are heuristics, and they are gameable: chopping every sentence in half improves the score while making the writing worse. Use them to find the passages worth rereading, not as a target.
Sources
Flesch R, “A new readability yardstick”, 1948. Kincaid JP et al., US Navy technical report, 1975. Gunning R, The Technique of Clear Writing, 1952. McLaughlin GH, “SMOG grading”, 1969. Coleman M & Liau TL, 1975. Syllable counts here are heuristic, so scores may differ by a fraction from dictionary-based implementations.