Readability grade: what the number measures and what it misses
A readability grade is a number that estimates how many years of schooling a reader needs to follow a text comfortably. It comes from a formula, not a reader. The common ones, Flesch-Kincaid Grade Level, Gunning Fog, SMOG and Coleman-Liau, all combine two things: how long the sentences are and how long the words are, measured in syllables or letters. Humanize 360 does not report a grade as one of its eleven signals, but several modes move it deliberately, and this page explains how to read the number without over-trusting it.
Two sentences make the point. "The board approved the budget." scores at a low grade: five short words. "The board, having considered the recommendations of the finance subcommittee, approved the consolidated budget." scores several grades higher: longer words, one long sentence. The second is not harder to understand for someone who reads reports every day. The formula cannot know that. It counts syllables.
What the formulas actually count
- Average sentence length in words.
- Average word length, in syllables (Flesch-Kincaid, Fog, SMOG) or letters (Coleman-Liau).
- For Fog, the share of words with three or more syllables.
- Nothing about vocabulary familiarity, sentence structure, coherence or whether the text is true.
Why it matters for writing that reads human
Readability and human texture are related but not the same. Machine text often sits at a steady middle grade, paragraph after paragraph, because its sentences and words are steadily medium. Human writing wanders: a technical paragraph, then a blunt one. So a stable grade across a whole document is itself a mild tell, and it lines up with the sentence-length variance and vocabulary spread signals on our detector. A low grade is not a goal. A grade that fits the reader is.
How Humanize 360 modes move it
- Simple mode shortens sentences and prefers common words, lowering the grade for general readers.
- Academic and Formal modes keep longer words and complex sentences, so the grade stays higher on purpose.
- Shorten mode cuts length without necessarily changing grade; Expand adds explanation and may raise or lower it.
- A custom style profile aims at your own sentence length, which sets the grade near where you naturally write.
Common misreadings
A high grade does not mean sophisticated thought; it means long words and sentences. A low grade does not mean dumbed down; most well-edited journalism sits low. Tools that report a "target grade" for essays or blog posts are applying a rule of thumb, not a standard. And no readability score is an AI detector: a model can be told to write at any grade, and a person can write at any grade. Look at the variance of the score across a piece, not the score itself.
Common questions
What readability grade should I aim for?
Whatever your reader expects. General web writing often sits around the level of a secondary-school reader; a journal article sits higher. Match the audience, then vary it within the piece.
Is readability one of the eleven Human Score signals?
Not directly. Sentence-length variance and vocabulary spread overlap with it, and a flat grade across a document tends to show up in those. Use the grammar checker or a word processor for the grade itself.
Does lowering readability grade make text read human?
No. Short uniform sentences read as machine too. Human texture comes from variation and specificity, not from a lower number.
Which formula is best?
They agree with each other reasonably well because they count the same things. Pick one and use it consistently rather than comparing numbers across formulas.