Flesch-Kincaid Readability Scores Explained: Formulas, Code, and SEO Writing

"The cat sat on the mat" scores differently from "The feline reposed upon the floor covering." Both say the same thing, but readability formulas quantify how hard each is to read. These scores power everything from Microsoft Word's readability checker to SEO content guidelines to US military technical manual standards.

This article explains the math behind the major readability formulas, with hand-calculated examples and working code.

Flesch Reading Ease: The Original (1948)

Rudolf Flesch's Reading Ease formula scores text from 0 to 100 (higher = easier):

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

Two inputs: average sentence length and average syllables per word. That's it. The constants (206.835, 1.015, 84.6) were derived empirically by Flesch from correlating these metrics with reading comprehension test scores.

Worked Example

Text: "The cat sat. The dog ran fast."

• Words: 7 (The, cat, sat, The, dog, ran, fast)
• Sentences: 2
• Syllables: The(1) + cat(1) + sat(1) + The(1) + dog(1) + ran(1) + fast(1) = 7

RE = 206.835 − 1.015 × (7/2) − 84.6 × (7/7)
   = 206.835 − 1.015 × 3.5 − 84.6 × 1.0
   = 206.835 − 3.5525 − 84.6
   = 118.68

Scores above 100 are possible for very simple text (the scale isn't strictly capped). Now a harder example:

Text: "The implementation utilizes sophisticated methodological frameworks."

• Words: 5, Sentences: 1
• Syllables: The(1) + im-ple-men-ta-tion(5) + u-ti-li-zes(4) + so-phis-ti-ca-ted(5) + meth-od-o-log-i-cal(6) + frame-works(2) = 23

RE = 206.835 − 1.015 × (5/1) − 84.6 × (23/5)
   = 206.835 − 5.075 − 84.6 × 4.6
   = 206.835 − 5.075 − 389.16
   = −187.4

Negative scores indicate extremely difficult text. The interpretation scale:

Score	Difficulty	Typical material
90–100	Very easy	5th grade, comics
80–90	Easy	6th grade, consumer ads
70–80	Fairly easy	7th grade
60–70	Standard	8th–9th grade, most web content
50–60	Fairly difficult	10th–12th grade
30–50	Difficult	College level
0–30	Very difficult	Academic papers, legal documents

SEO sweet spot: 60–70. Most successful web content targets this range — accessible to a broad audience without sounding dumbed down.

Flesch-Kincaid Grade Level: The US Navy Version (1975)

The US Navy needed to ensure technical manuals matched sailors' reading levels. J. Peter Kincaid recalibrated Flesch's formula to output a US grade level instead of a 0–100 score:

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

Using our examples:

Simple text ("The cat sat. The dog ran fast."):

FKGL = 0.39 × 3.5 + 11.8 × 1.0 − 15.59
     = 1.365 + 11.8 − 15.59
     = −2.43  →  effectively grade 0 (very easy)

Hard text ("The implementation utilizes..."):

FKGL = 0.39 × 5.0 + 11.8 × 4.6 − 15.59
     = 1.95 + 54.28 − 15.59
     = 40.64  →  grade 40+ (post-graduate+)

A grade level of 8.0 means an 8th grader (age ~13–14) should comprehend it. Most web content guidelines recommend grade 7–9.

Gunning Fog Index

Robert Gunning's Fog Index (1952) takes a different approach — it counts complex words (3+ syllables, excluding proper nouns, compound words, and common suffixes):

Fog = 0.4 × ((words/sentences) + 100 × (complex_words/words))

Example: "The cat sat on the mat. It was happy."

• Words: 10, Sentences: 2, Complex words: 0
• Fog = 0.4 × ((10/2) + 100 × (0/10)) = 0.4 × 5 = 2.0

The Fog Index roughly corresponds to years of education needed. A Fog of 12 = high school senior level. The Economist averages ~11; academic papers often exceed 16.

SMOG Index

McLaughlin's SMOG (Simple Measure of Gobbledygook, 1969) focuses purely on polysyllabic words:

SMOG = 1.043 × √(polysyllables × (30/sentences)) + 3.1291

It's specifically calibrated for 30-sentence samples. Fewer sentences? The formula still works but accuracy drops. SMOG tends to score 1–2 grades higher than Flesch-Kincaid for the same text — it's the "strict" grader.

The Hard Part: Counting Syllables

All these formulas depend on syllable counts, but English syllabification is notoriously irregular. Production implementations use heuristic algorithms:

import re

def count_syllables(word: str) -> int:
    """Heuristic syllable counter. Accurate ~90% of the time."""
    word = word.lower().strip()
    if not word:
        return 0
    if len(word) <= 3:
        return 1

    # Remove silent 'e' at end
    word = re.sub(r'e$', '', word)
    # Remove 'es', 'ed' endings (usually silent)
    word = re.sub(r'(es|ed)$', '', word)

    # Count vowel groups
    vowels = re.findall(r'[aeiouy]+', word)
    count = len(vowels)

    # Adjustments
    if word.endswith('le') and len(word) > 2 and word[-3] not in 'aeiouy':
        count += 1  # 'table' = 2, not 1

    return max(1, count)

# Test
for w in ["cat", "implementation", "the", "beautiful", "queue"]:
    print(f"{w}: {count_syllables(w)}")
# cat: 1, implementation: 5, the: 1, beautiful: 3, queue: 1

Edge cases that break naive counters: "queue" (1 syllable, 4 vowels), "rhythm" (2 syllables, 0 standard vowels — the 'y' counts), "business" (2 syllables, not 3). Dictionary-based approaches (like the CMU Pronouncing Dictionary) are more accurate but require a 130,000-word lookup table.

Complete Readability Calculator

function fleschKincaid(text) {
  const sentences = text.split(/[.!?]+/).filter(s => s.trim().length > 0);
  const words = text.match(/\b[\w']+\b/g) || [];
  const syllables = words.reduce((sum, w) => sum + countSyllables(w), 0);

  const wps = words.length / Math.max(1, sentences.length);  // words per sentence
  const spw = syllables / Math.max(1, words.length);          // syllables per word

  const readingEase = 206.835 - 1.015 * wps - 84.6 * spw;
  const gradeLevel = 0.39 * wps + 11.8 * spw - 15.59;

  return {
    words: words.length,
    sentences: sentences.length,
    syllables,
    readingEase: Math.round(readingEase * 10) / 10,
    gradeLevel: Math.round(gradeLevel * 10) / 10,
  };
}

console.log(fleschKincaid("The cat sat on the mat. It was a happy cat."));
// { words: 11, sentences: 2, syllables: 12,
//   readingEase: 109.3, gradeLevel: -0.6 }

Limitations (What Scores Don't Tell You)

Readability formulas are useful but crude. They don't measure:

• Coherence: "Colorless green ideas sleep furiously" scores as easy (short words, short sentences) but is nonsense. (Chomsky's famous example.)
• Domain knowledge: "The mitochondria is the powerhouse of the cell" is "difficult" by syllable count but trivial for a biology student.
• Layout and formatting: bullet points, headings, and white space dramatically affect real readability but don't change the score.
• Cultural context: idioms and references that are easy for one audience are opaque to another.

Use scores as a rough gauge, not a target to game. Writing to hit "grade 8.0" by mechanically shortening sentences produces stilted prose. Better: write naturally, check the score, and revise if it's wildly off from your target audience.

SEO Writing: Practical Guidelines

For web content targeting a general audience:

1. Aim for Flesch Reading Ease 60–70 (grade 7–9). This covers the broadest audience without sounding simplistic.
2. Sentences under 20 words on average. Vary length for rhythm — all short sentences sound robotic.
3. Prefer Anglo-Saxon words over Latinate ones: "use" not "utilize," "help" not "facilitate," "start" not "commence."
4. One idea per sentence. If you need a semicolon, consider splitting.
5. Active voice tends to produce shorter, clearer sentences (though passive has its place).

Google doesn't directly use readability scores as a ranking factor, but user engagement metrics (bounce rate, time on page, scroll depth) correlate strongly with readability. Content people can actually read performs better.

Try It

Want to check your content's readability right now? DailyToolbox's free readability analyzer runs entirely in your browser: paste your text and get instant Flesch-Kincaid grade level, reading ease score, word count, sentence count, and estimated reading time. Pair it with the word counter for quick content audits.

👉 Try the Readability Analyzer →

FAQ

What's a good Flesch-Kincaid score for a blog post?

Grade 7–9 (Reading Ease 60–70) for general audiences. Technical blogs can go higher (grade 10–12) if the audience is specialized, but even expert readers prefer clear writing. The New York Times averages around grade 10; Reader's Digest targets grade 8.

Why do different tools give different readability scores?

Syllable counting is heuristic. Different implementations handle edge cases (silent 'e', 'y' as vowel, compound words) differently. Sentence splitting also varies (how to handle "Dr.", "e.g.", decimal numbers). Expect ±1 grade level variation between tools — that's normal.

Do readability scores work for non-English text?

Not reliably. The formulas were calibrated on English. Some adaptations exist (Fernández Huerta for Spanish, LIX for Swedish/Danish), but you can't just run English Flesch-Kincaid on French text and trust the result. Syllable structures and sentence patterns differ fundamentally across languages.

Should I dumb down my writing to hit a target score?

No. Write for your audience first, then check the score as a sanity check. If you're writing for PhD researchers and score grade 16, that's fine. If you're writing a consumer guide and score grade 16, simplify. The score is a diagnostic tool, not a writing instructor.