What to measure instead of a readability score
A reading ease score is not a measure of plain language. It counts two things: average sentence length, and average syllables per word. Then it reports the result as a school grade. That is useful information. It is not the same as whether a person can read your page and do the thing they came to do.
The gap between those two is where a lot of government content quietly fails. The confusion between plain language and a readability score gets built into style guides, content KPIs, and accessibility sign-offs. Once it is in there, it is hard to see.
Here is the problem in one paragraph. Read this:
You may be eligible. Eligibility is subject to criteria. Criteria are set out below. A determination will be made. You will be notified of the outcome.
Short sentences. Common words. It scores well, at around a year 7 grade. It also tells the reader nothing they can act on. They do not know if they qualify, what the criteria are, what to do next, or when they will hear back. The maths passed. The person did not.
What a readability formula actually measures
Flesch-Kincaid, and the reading-ease scores built on the same method, measure sentence length and syllable count. That is the whole input. The formula cannot see:
whether a term is defined
whether the steps are in the order the reader needs them
whether the heading matches the task
whether the one sentence that answers the reader's question is even on the page
This is not a flaw in the formula. It is doing exactly what it was designed to do decades ago. A readability score is a smoke alarm, not a certificate. It can tell you a sentence is probably too long. It cannot tell you the content works.
What plain language actually requires
Plain language has a definition, and it is not about syllables. AS ISO 24495-1:2023, now an Australian Standard, sets that definition: communication is plain language when the wording, structure, and design let the intended reader find what they need, understand it, and use it.
Find, understand, use. Every one of those is about the reader's task, not the writer's sentence length. A page can score at year 7 and fail all three. A page can carry a slightly longer sentence and pass all three, because the reader gets the answer and knows what to do.
The score checks the sentence. Plain language checks the reader.
| Readability score | Plain language | |
|---|---|---|
| What it measures | Sentence length and syllables per word | Whether the reader can find, understand, and use the content |
| What it is based on | A readability formula | AS ISO 24495-1:2023, an Australian Standard |
| The question it answers | Is the writing mechanically simple? | Can the reader act on it? |
| Best used as | A quick check for bloated sentences | The actual measure of success |
Why this matters for accessibility, not just style
The reading-level target most agencies cite comes from accessibility. WCAG success criterion 3.1.5 asks for content at a lower secondary reading level, and the Style Manual's year 7 recommendation traces back to it. That lineage is real, and it is where the measure gets its authority.
But WCAG 3.1.5 is a Level AAA criterion, and the way to satisfy it is not to hit a number. It is to provide content, or a supplementary version, that a lower secondary reader can understand. A readability score is evidence you might be close. It is not conformance, and treating it as conformance means content can clear an accessibility check while still shutting people out.
A passing score is not proof of access. It is a guess that has not been tested.
What to measure instead
Keep the readability tool. It is a fast, free way to catch a bloated sentence. Just stop letting it sign off the work.
The measure that matters is whether the reader can act. You can test that cheaply:
Read the page aloud. The place you stumble is the place they stumble.
Ask five people from the actual audience to read it once, then tell you what they would do next. If they cannot, the content failed, whatever the score said.
Check the page against find, understand, use. Can the reader find the answer, understand it, and use it without help?
In practice, the teams that get this right are not the ones with the best readability dashboards. They are the ones who put draft content in front of a real person before it publishes, and who treat "they got it" as the pass mark.
So the question to put to any content that just cleared a readability check: could the person it is for read it once and know what to do next? If nobody has asked a real reader that, you have measured the writing. You have not measured the reading.
Photo by Dmitry Ratushny on Unsplash
Did this land?
I write these to be used, not just read. Tell me if this wouldn't work in your organisation, and why.