Understanding Readability Scores: Flesch-Kincaid Explained

Every readability formula in common use rests on the same two observations: long sentences are harder than short ones, and long words are harder than short ones. That is it. Everything else — the constants, the scales, the grade levels — is calibration on top of those two variables.

Understanding that is what lets you use a readability score properly, which means treating it as a fast structural diagnostic rather than a grade on your prose. This post covers what the formulas actually compute, how to read the numbers, and where they mislead. For the wider set of text measurements these scores sit alongside, see The Complete Guide to Counting Words and Characters.

The Two Flesch Scores

Rudolf Flesch published the Reading Ease formula in 1948. J. Peter Kincaid adapted it for the US Navy in 1975 to express results as a school grade level. Both are in the public domain, which is why they appear in Microsoft Word, Google Docs, Yoast, and effectively every readability tool built since.

Flesch Reading Ease:

206.835 − 1.015 × (words / sentences) − 84.6 × (syllables / words)

Flesch-Kincaid Grade Level:

0.39 × (words / sentences) + 11.8 × (syllables / words) − 15.59

Note the coefficients. In Reading Ease, syllables per word carries an 84.6 weight against 1.015 for sentence length — vocabulary dominates. Swapping "utilise" for "use" moves the needle far more than trimming a sentence by two words.

Reading the Reading Ease Scale

Score	Difficulty	Rough grade	Typical example
90–100	Very easy	5th grade	Children's books
80–90	Easy	6th grade	Consumer advertising
70–80	Fairly easy	7th grade	Popular fiction
60–70	Plain English	8th–9th grade	Newspapers, good web copy
50–60	Fairly difficult	10th–12th grade	Trade publications
30–50	Difficult	College	Academic articles
0–30	Very confusing	College graduate	Legal and medical texts

The scale is not strictly bounded. Text made of very long words in very long sentences can score below zero, and a document of three-word sentences using only monosyllables can exceed 100. Both are pathological cases you will only see in test data.

What to Target

The right score depends entirely on audience, not on some universal ideal.

Content type	Reading Ease	Grade level
Marketing and landing pages	70–80	6–7
Blog posts and general web content	60–70	8–9
Email newsletters	65–75	7–8
Product documentation	50–60	10–11
B2B and trade writing	45–60	10–12
Academic papers	30–45	13+
Legal and medical	below 30	15+

Writing for a specialist audience at grade 6 is not a virtue; it reads as condescending and forces you to strip the precise terminology those readers rely on. The goal is matching, not maximising.

One genuinely universal benchmark: public-facing health, safety, and government information should target grade 6 to 8, because that is where general-population comprehension holds up under stress or time pressure.

The Other Formulas You Will Meet

Most tools report several scores. They differ mainly in which proxy they use for word difficulty.

Gunning Fog Index — 0.4 × [(words/sentences) + 100 × (complex words/words)], where a complex word has three or more syllables. Outputs a grade level. Fog is stricter than Flesch-Kincaid on jargon-heavy writing because polysyllabic words are counted directly rather than averaged.

SMOG Index — based on the count of polysyllabic words in a 30-sentence sample. Widely used in healthcare, where it is considered the more conservative choice for patient materials.

Coleman-Liau — 0.0588L − 0.296S − 15.8, where L is letters per 100 words and S is sentences per 100 words. It counts characters instead of syllables, which makes it far more reliable for software, since syllable counting in English is genuinely hard.

Automated Readability Index — 4.71 × (characters/words) + 0.5 × (words/sentences) − 21.43. Also character-based, and usually within a grade of Coleman-Liau.

When formulas disagree by more than two grades, the usual cause is a text with either unusual syllable patterns (many short polysyllables like "area" or "video") or heavy proper-noun density. The Readability Score (Flesch-Kincaid) shows several formulas together, and the spread between them is itself informative — tight agreement means a confident reading.

Why Syllable Counting Is the Weak Link

Flesch and Gunning Fog both need a syllable count, and no algorithm gets English syllables exactly right. Software approximates by counting vowel groups and applying exception rules: subtract a silent trailing "e", never return zero, treat "le" endings specially. That handles most words and then fails on "queue" (one syllable, five vowels), "business" (two, not three), and every irregular loanword.

The practical effect is a margin of error around half a grade level on ordinary prose, more on text dense with names or technical terms. It is a reason to read a Flesch-Kincaid grade of 9.2 as "roughly ninth grade" rather than a measurement.

The Real Limitation

A readability formula cannot read. It has no access to whether your argument follows, whether the jargon is defined, or whether the paragraph order makes sense. Two consequences follow.

Nonsense scores well. Randomly shuffle the sentences of a well-written article and the readability score does not change at all — the words and sentence boundaries are identical. That single fact should calibrate how much weight the number deserves.

Optimising directly makes writing worse. Splitting every sentence at the nearest conjunction raises the score and produces choppy, monotonous prose. Varied sentence rhythm — a 6-word sentence after a 28-word one — is good writing that averages out to a mediocre score.

Use the score the way you would use a smoke alarm: it is excellent at telling you when something is wrong, and useless as a guide to cooking. A Reading Ease of 22 on a blog post means go look at your sentences. A Reading Ease of 64 means nothing needs your attention, not that the piece is good.

A Practical Editing Pass

Run this on a finished draft, not while writing.

1. Get the baseline from the Readability Score (Flesch-Kincaid) and compare it to the target for your content type.
2. If the score is low, check average sentence length first — the Text Statistics Summary gives it directly. Above 25 words per sentence is the single most common cause, and the easiest to fix.
3. Next, look at your longest words. Replacing five instances of "implementation" with "rollout" moves the score more than restructuring three paragraphs, because of that 84.6 coefficient.
4. For pages where search visibility matters, the Flesch-Kincaid SEO Readability Checker frames the same measurements against content-marketing benchmarks rather than academic ones.
5. Re-score once, then stop. A second optimisation round is where prose starts getting damaged.

Everything runs client-side, so an unpublished draft stays on your machine. And if the piece feels bloated as well as dense, length is a separate question — Word Count Targets for Essays, Blogs and Emails covers what each format should actually weigh.