Readability scores attempt to measure how easy a piece of text is to read. They use mathematical formulas based on word length, sentence length, and syllable count to estimate the reading level required to understand the text. I find them useful as a rough editing guide but dangerously misleading when treated as quality metrics. A readability score can tell you that your sentences are long. It cannot tell you whether your writing is clear, accurate, or worth reading.
Here is how the major readability formulas work, what scores to aim for, and where they fall short.
Flesch-Kincaid Reading Ease
The Flesch Reading Ease score rates text on a scale from 0 to 100. Higher scores mean easier reading. The formula uses average sentence length (words per sentence) and average syllables per word. Short sentences with short words score high. Long sentences with multisyllabic words score low.
Score ranges: 90 to 100 is easily understood by an 11-year-old. 60 to 70 is easily understood by 13 to 15-year-old students. 30 to 50 suits college-level readers. Below 30 is best understood by university graduates. Most web content should aim for 60 to 70, which matches the reading level of a general adult audience scanning content on a screen.
The related Flesch-Kincaid Grade Level formula converts the same inputs into a US school grade level. A score of 8 means the text is appropriate for an eighth-grade student. Most successful blogs and news sites write at a grade 6 to 8 level, not because their readers lack education but because simpler text is faster to scan and easier to absorb when reading online.
Gunning Fog Index
The Gunning Fog Index estimates the years of formal education needed to understand a text on first reading. It factors in sentence length and the percentage of complex words, defined as words with three or more syllables. A Fog Index of 12 means the text requires roughly 12 years of education, equivalent to a high school senior.
The Fog Index tends to penalise technical writing more heavily than Flesch-Kincaid because technical terms are often long and multisyllabic regardless of whether the audience finds them difficult. "Database" has three syllables and would be flagged as complex, but any developer reads it without effort. This makes the Fog Index less useful for specialised content where the audience is familiar with domain terminology.
For general web content, aim for a Fog Index between 7 and 12. Newspapers like The Wall Street Journal typically score around 11. Tabloids score around 6 to 8. Children's books score below 6.
Coleman-Liau Index
The Coleman-Liau Index is unusual because it uses character count instead of syllable count. It measures average word length in characters and average sentence length in words. The output is a US grade level. This formula is easier to calculate programmatically because counting characters is simpler and more reliable than counting syllables, which requires language-specific rules for handling silent letters, diphthongs, and irregular words.
I find the Coleman-Liau Index gives more consistent results than syllable-based formulas for the same reason: character counting is deterministic while syllable counting involves judgment calls. Is "fire" one syllable or two? Is "comfortable" three syllables or four? These ambiguities do not affect character-based formulas.
SMOG Index
SMOG stands for Simple Measure of Gobbledygook. It estimates the years of education needed to understand a text by counting polysyllabic words, those with three or more syllables, in a sample of 30 sentences. The formula takes the square root of the polysyllabic word count multiplied by a constant.
SMOG was designed for healthcare communications and is widely used in that field. It tends to produce higher grade level estimates than other formulas, making it a conservative measure. If your text passes a SMOG readability check, it will almost certainly pass checks using other formulas too. Health literacy guidelines in the US and Australia recommend targeting a SMOG grade of 6 to 8 for patient-facing materials.
What Readability Scores Miss
Every readability formula reduces complex human reading to two or three numeric inputs. They measure surface features of text, not comprehension difficulty. A sentence using simple, short words can be confusing if the ideas are poorly organised. A sentence using longer words can be perfectly clear if the structure guides the reader logically.
Readability scores cannot evaluate coherence, logical flow, accuracy, or whether the content actually answers the reader's question. I have seen articles with perfect readability scores that say nothing useful, and technical documentation with poor scores that communicates complex ideas with impressive clarity.
The formulas also penalise necessary complexity. If you are writing about cardiovascular disease, you cannot avoid multisyllabic medical terms without sacrificing precision. Replacing "atherosclerosis" with "blocked arteries" improves the readability score but changes the meaning. The right approach is to use the precise term and explain it, not to dumb down the vocabulary to chase a number.
Using Scores Practically
I check readability scores after writing, never during. Writing to hit a target score produces stilted, unnatural text because you start optimising for the formula instead of the reader. Write naturally first. Check the score. If it is significantly outside your target range, look at why.
Usually the problem is one of two things: sentences that are too long, or unnecessary jargon. Long sentences are easy to fix by splitting at natural breakpoints. Jargon is easy to fix by asking whether your specific audience needs the technical term or whether a simpler word communicates the same meaning. Make these targeted edits, recheck the score, and stop when you are in range.
The word counter gives you sentence count and average sentence length, which are the two factors that most strongly influence readability scores. If your average sentence length is over 20 words, your readability scores will be higher than ideal for web content. Getting that average down to 15 to 18 words usually brings all the readability formulas into acceptable ranges simultaneously.