What is a readability score, and should you trust it?
A formula estimating how hard a text is to read, almost always by combining sentence length with word length or syllable count, and outputting a grade level or a score. They are genuinely useful as a rough signal and systematically misleading when treated as a target.
The common ones:
Flesch Reading Ease, scoring 0–100 with higher being easier; plain-English guidance often suggests aiming for 60 or above for a general audience.
Flesch–Kincaid Grade Level, expressing the same inputs as a US school grade.
Gunning Fog, which counts "complex" words by syllable count.
SMOG, widely used for health information.
What they actually measure. Two proxies, and only two: how long your sentences are, and how long your words are. Everything else a reader struggles with is invisible to them.
Why that makes them easy to game and easy to misuse:
They cannot detect nonsense. Text made of short words in short sentences scores beautifully whether or not it means anything.
Familiarity beats length. "Cardiomyopathy" is long and clear to a cardiologist; "we will be in touch in due course" is short, simple and tells the reader nothing about when.
They ignore structure entirely — headings, ordering, paragraphing, signposting — which affects comprehension more than syllable counts do.
They ignore the passive voice, nominalisation and ambiguity, which are the real sources of impenetrable prose.
Chopping sentences to hit a target can reduce clarity by removing the connectives that showed how ideas related.
How to use them properly:
As a smoke alarm, not a thermostat. A bad score means look again; a good score proves nothing.
Check what is driving it — one enormous sentence or a genuine pattern.
Test with actual readers, which is the only real measure.
Follow plain-language practice instead: one idea per sentence, active voice, concrete verbs, the reader's vocabulary, and the main point first.