How to Check Readability in Word — and Why Easy Reading Doesn't Mean Human Writing

How Do You Turn On Readability Statistics in Word?

Word does not show readability scores automatically — you have to enable the setting once, and then it appears every time you run a spell check. Once it's on, checking readability in Word becomes part of the normal spell-check flow rather than a separate task. The option lives in the proofing settings rather than the ribbon, which is why many long-time Word users have never seen it.

1. Open Word and go to File > Options (on Mac: Word > Preferences > Spelling & Grammar)
2. Click Proofing in the left sidebar
3. Under 'When correcting spelling and grammar in Word,' check the box labeled 'Show readability statistics'
4. Click OK to save the setting
5. Run a spell check on your document (Review > Spelling & Grammar, or press F7)
6. After Word finishes checking spelling and grammar, a Readability Statistics dialog box will open automatically

What Do the Readability Statistics in Word Actually Mean?

The Readability Statistics dialog shows three groups of information: basic counts (words, characters, paragraphs, sentences per paragraph, words per sentence), a passive-voice percentage, and two Flesch-based scores. The two scores are the ones people usually mean when they ask how to check readability in Word, and they measure the same underlying thing from opposite directions.

1. Flesch Reading Ease: scored 0–100, where higher means easier to read. Scores of 60–70 are considered plain English understood by most 13-to-15-year-olds; below 30 is roughly college-graduate level; above 90 is very easy
2. Flesch-Kincaid Grade Level: converts the same sentence-length and syllable-count data into a U.S. school grade level, so a score of 8.0 suggests the text is readable by an eighth grader
3. Both formulas rely on only two inputs: average sentence length and average syllables per word — they do not evaluate vocabulary difficulty, logical structure, or whether the content makes sense
4. Passive sentences percentage: a separate metric showing how much of the document uses passive voice, which Word flags because passive constructions typically add words without adding clarity

Readability formulas are a proxy for reading effort, calculated from sentence length and syllable count. They say nothing about accuracy, tone, or whether a sentence actually communicates the intended idea.

What Readability Score Should You Be Aiming For?

The right target depends entirely on the audience and format, not on a universal 'good' number. A grade level appropriate for a legal brief would be a problem in a product landing page, and vice versa.

1. General web content and marketing copy: aim for Flesch Reading Ease 60–70 (roughly 7th–8th grade), which most adult readers process quickly on a screen
2. Newspaper and mainstream journalism style writing: Reading Ease around 60–65, similar to outlets that write for a broad public audience
3. Academic or technical documentation: Reading Ease can reasonably drop to 30–50, since specialized vocabulary is unavoidable and the audience already has domain knowledge
4. Instructions, help articles, and onboarding text: aim as high as 70–80 — the goal is that a reader under time pressure understands the step on the first read
5. Emails and internal memos: no fixed target, but a sudden jump in grade level compared with your usual writing is worth a second look before sending

Why Does Word Say My Writing Is Hard to Read Even When It's Correct?

A high Flesch-Kincaid grade level usually traces back to a small number of habits rather than a fundamental writing problem, and each one is easy to isolate in the Readability Statistics numbers.

1. Long sentences with multiple clauses joined by 'which,' 'that,' and 'although' — splitting a 35-word sentence into two shorter ones lowers the grade level without losing meaning
2. Multisyllabic vocabulary where a shorter word would do — 'utilize' versus 'use,' 'facilitate' versus 'help' — the formula counts syllables directly, so word choice has an outsized effect
3. Heavy passive voice, which tends to lengthen sentences ('the report was reviewed by the committee' versus 'the committee reviewed the report')
4. Nested qualifications and parenthetical asides that push the average words-per-sentence count up even when each individual clause is simple
5. Dense technical or legal terminology that is genuinely necessary for the subject — in this case a high grade level reflects the topic, not a writing flaw, and simplifying further may cost precision

Does a Good Readability Score Mean Word Detection Tools Will Pass It?

No — readability and AI detection measure completely different properties of a text, and it's a common mix-up once writers start editing for both. Flesch-Kincaid only looks at sentence length and syllable count; it has no way to tell who or what produced the words. AI detection tools instead look at perplexity (how predictable each word choice is) and burstiness (how much sentence length and structure vary across a document). A short, simple, low-grade-level sentence can still read as highly predictable to a detector, and a long, complex sentence can still read as distinctly human, because the two measurements are built from different signals entirely. In practice, editing purely for a lower readability score can sometimes push a document in the wrong direction for AI detection — smoothing every sentence to a similar, easy-to-read length is exactly the kind of low-variation pattern that raises AI-likeness scores, since it removes the natural burstiness of human writing. The two goals aren't opposed, but they aren't the same lever either, so it's worth checking both separately rather than assuming one improves the other.

A text can score 70 on Flesch Reading Ease and still read as highly AI-like, because readability formulas measure sentence and word length, not the statistical predictability that AI detectors are built to flag.

How Should You Check Your Writing After Editing for Readability?

Knowing how to check readability in Word is only the first step. Once a document passes your target readability score, it's worth a separate check for how natural and varied the writing actually reads — especially for content that will pass through institutional AI screening, editorial review, or a hiring pipeline. NotGPT's AI text detector analyzes a pasted document for perplexity and burstiness and highlights the specific sentences that read as most predictable, which is often the same set of sentences that got flattened during a readability pass. If a section comes back flagged, running it through NotGPT's Humanize tool at a light intensity can restore natural sentence-length variation without reintroducing the long, dense sentences that hurt the readability score in the first place. Checking both — Word's readability statistics and an AI-likeness score — before submitting or publishing gives a fuller picture than optimizing for either one alone.