Find out how hard your writing is to read

Readability formulas answer one narrow question: roughly how much schooling a reader needs to get through a sentence on the first attempt. They are used to check that a public notice, a consent form, a policy document or a landing page is not written for a smaller audience than it was meant for. Plain-English requirements in government and healthcare are usually written as a target on one of these scales.

Six are reported here rather than one, and they are not averaged into a single figure. Averaging them would be inventing a seventh measure while borrowing the credibility of the six that have citations behind them. They disagree with each other on purpose: Flesch-Kincaid and Automated Readability weigh sentence length heavily, Gunning Fog and SMOG count long words, and Coleman-Liau counts letters rather than syllables and so is unaffected by how the syllable estimate is made. Where they agree you can trust the direction. Where they scatter, the text is probably uneven rather than uniformly hard.

All six are estimates, and it is worth being precise about why. Every one except Coleman-Liau depends on counting syllables, and English syllables cannot be counted exactly by rule. “Business” is two syllables and “busyness” is three, “create” is two and “beautiful” is three, and no set of spelling rules separates those cases without a pronunciation dictionary. The counter here handles the common patterns and misses a class of words where two vowels straddle a syllable boundary. That is why a difference of a grade or less between two drafts means nothing.

The part that is exact is underneath the grades: the longest sentences in the text, with their word counts. Sentence length is the largest single term in most of these formulas, it is the thing you can actually change, and finding the three worst sentences in a long document by eye is the tedious part of editing. Shortening them moves every score at once, and unlike the grades, the count of words in a sentence is not an estimate of anything.

SMOG is reported even on short passages, with a note when the text is under the thirty sentences its definition assumes. Hiding it would be tidier and less honest: a number outside its intended range is still information, provided the page says so.

The measurement runs in this tab. That matters more than it sounds for this particular tool, because the drafts people most want to check for plain English are the ones they are least willing to paste into somebody else’s server: patient letters, benefit decisions, redundancy notices, terms that are about to be published under a company’s name.