In short
A readability score is a number produced by a formula that estimates how easy a passage is to read, based almost entirely on average sentence length and average word length or syllable count; Flesch Reading Ease, Flesch-Kincaid Grade Level, Gunning Fog and SMOG are the common ones. Most web copy for a general audience should score around a US grade level of 7 to 9, and plain-language guidance aims lower. The scores are useful for catching over-long sentences and jargon, and useless for judging clarity, structure or accuracy, so use them as a check after writing rather than a target while writing.
What the formulas measure
Every mainstream readability formula uses two inputs: how long the sentences are and how long the words are. Long sentences and multi-syllable words are harder to process, on average, so the formulas combine the two averages into a score.
Flesch Reading Ease produces a number from 0 to 100, higher being easier: 60 to 70 is plain English readable by most adults; 30 and below is academic.
Flesch-Kincaid Grade Level converts the same inputs into a US school grade: a score of 8 means an average eighth-grader, roughly a 13-year-old, could read it.
Gunning Fog estimates years of education needed, weighting words of three or more syllables.
SMOG does similarly with a different formula and is often used for health materials.
Automated Readability Index and Coleman-Liau use character counts instead of syllables, which makes them easier to compute.
They correlate with each other and with reading difficulty in broad terms. They were developed for print in the mid-twentieth century, mostly for military and educational materials, and they have been used ever since because they are cheap to calculate and roughly right.
Sensible targets
| Content | Flesch-Kincaid grade | Flesch Reading Ease |
|---|---|---|
| General web copy, marketing pages | 7–9 | 60–70 |
| Public services, health, legal notices for the public | 6–8 | 65–75 |
| Product interface text, error messages | 5–7 | 70–80 |
| Professional and technical audiences | 9–12 | 45–60 |
| Academic and specialist | 12+ | Below 45 |
The general-audience target is lower than most people expect. Average reading ability in the UK and US is around grade 8 to 9, many readers are below it, most are reading on phones with divided attention, and even expert readers prefer simpler text when the content is not their specialism. Writing at grade 8 is not condescension; it is respect for attention.
What the scores miss
They cannot tell whether the words are in a sensible order, whether the argument holds, whether the jargon is necessary for the audience, or whether the structure helps. A passage of short, simple sentences can be incoherent and score beautifully. A well-organised technical explanation with necessary long words can score badly and be perfectly clear to its readers. Scores ignore headings, lists, layout, images and all the visual factors that determine whether text is actually read.
They are also easy to game. Chopping sentences in half and swapping precise words for vague short ones improves the score and worsens the writing. Anyone who has edited to a target number has produced at least one paragraph that reads like a telegram.
Tools
Microsoft Word and Google Docs report Flesch scores. Hemingway Editor highlights hard sentences and gives a grade. Grammarly, Readable and various free calculators report the full set. Content management systems and SEO plugins increasingly include a score in the editor. Search engines do not use readability scores as a ranking signal, though clearer text tends to perform better for the ordinary reason that people read it.
How to use them
After writing, not during. Write for the reader, then run the score. If it is high for the audience, look for the causes rather than the number: sentences over 25 words, paragraphs that do not break, terms that could be replaced, passive constructions that hide the actor.
As a comparison. Score a page that performs well and one that does not; the difference is often instructive. Score a draft against the previous version to check an edit made it simpler rather than just shorter.
As a flag for specific sentences. Tools that highlight the hardest sentences are more useful than the aggregate score, because they point at the fix.
With judgement about the audience. A page for tax professionals is allowed to say "amortisation"; a page for first-time buyers should explain it or avoid it. The score does not know which page it is reading.
Alongside real readers. Five people from the audience reading a page aloud will find every problem the score missed and confirm every one it found. That is plain language practice, and it is the actual test.
The practical rule
Aim for a grade 7 to 9 on general pages, lower on interface text, and treat any score above the target as a prompt to find the long sentences and the unnecessary words. Never edit to the number. The goal is a page a distracted person on a phone can follow on the first read, and the score is a rough proxy for one part of that. If you would like a site's key pages checked for readability by both the formula and a reader, book a call.
Common questions
What is a good readability score for a website?
For a general audience, a Flesch-Kincaid grade level of about 7 to 9, or a Flesch Reading Ease of 60 to 70. Interface text and error messages should be simpler, around grade 5 to 7. Content for professional and technical audiences can sit at grade 9 to 12. Average adult reading ability is around grade 8, and most people prefer simpler text than they can manage.
What does the Flesch-Kincaid grade level mean?
The US school grade at which an average student could read the text, computed from average sentence length and average syllables per word. A grade of 8 corresponds roughly to a 13-year-old reader. It estimates surface difficulty only and says nothing about whether the text is clear, well organised or accurate.
Do readability scores affect SEO?
Not directly; search engines do not use them as a ranking signal. Clearer text tends to perform better because people read it, stay longer and act on it, and because direct, plainly written answers are what featured snippets and AI assistants extract. Treat readability as a quality practice rather than an optimisation.
How do I improve my readability score?
Find the causes rather than editing to the number: split sentences over about 25 words, break long paragraphs, replace multi-syllable words with common ones where meaning allows, use the active voice, and cut words that do nothing. Tools that highlight the hardest sentences point at the fixes. Chopping sentences arbitrarily raises the score and ruins the prose.
Are readability formulas accurate?
Roughly, for surface difficulty. They correlate with reading effort in broad terms but measure only sentence and word length, ignoring meaning, structure, layout, jargon appropriate to the audience and everything else that determines whether text is understood. Use them as a check for over-long sentences and unnecessary complexity, and test with real readers for the rest.
