What it measures
The checker counts words, sentences, and estimated syllables. From those three ingredients it calculates average sentence length, average syllables per word, and four standard readability formulas. It also lists every sentence longer than twenty-five words and the most frequent words with three or more syllables.
Those lists are often more useful than the headline score. A grade level tells you that a text is dense; a list tells you where. Maybe one forty-word sentence carries three ideas. Maybe the same technical term appears eleven times and could be defined once. Maybe the text is short and simple but the vocabulary assumes a specialist. The score points at a symptom; the list points at a sentence.
The four scores
Flesch Reading Ease is scored roughly from 0 to 100, with higher numbers easier to read. The bands are approximate: 90–100 is very easy, 80–90 easy, 70–80 fairly easy, 60–70 plain English, 50–60 fairly difficult, 30–50 difficult, and below 30 very difficult. The formula rewards short sentences and short words. It does not know whether a short sentence is clear or merely vague.
Flesch–Kincaid Grade converts similar inputs into a U.S. school-grade estimate. A score of 8 suggests a reader around eighth grade; a score of 13 suggests college-level reading. The number is a convention, not a measurement of a person. It says nothing about whether the content is appropriate, accurate, or interesting.
Gunning Fog combines average sentence length with the percentage of words that have three or more syllables. It tends to punish nominalisations — “implementation”, “documentation”, “authorisation” — which is useful when a draft has turned actions into abstract nouns. SMOG also focuses on complex words, but its estimate is unreliable on short texts. The tool labels it as an estimate and shows the score anyway, because a rough signal is still a signal.
The formulas overlap, and they can disagree. That is not a bug. Each one is a different simplification of a complicated reading experience. Pick the one that matches your context, keep it constant across drafts, and use the underlying lists to decide what to change.
Long sentences and complex words
A sentence longer than twenty-five words is not automatically bad. Academic prose, legal definitions, and literary sentences can carry length well. But length is a useful place to look first, because a long sentence often contains more than one idea and more than one possible subject. The checker lists those sentences with their word counts so you can decide which ones deserve a split.
Complex words are not automatically bad either. “Reimbursement” may be the precise term in a policy; “hypertension” may be the term a patient needs to recognise. The question is whether the reader already knows the word or can learn it from context. A list of repeated complex words is a prompt to check, not an instruction to replace every three-syllable word with a shorter one.
Editing without gaming
It is easy to raise a readability score by chopping every sentence into fragments. The result can be worse: choppy, repetitive, and harder to follow. Use the score to find friction, not as a target to optimise. A better edit usually does one of four things:
- Split at a clause boundary. Put the condition, the action, and the result in separate sentences when they are genuinely separate ideas.
- Name the actor. “The committee approved the change” is easier than “approval of the change was obtained”. The second version hides who did what.
- Turn nominalisations back into verbs. “We implemented the policy” is shorter and clearer than “the implementation of the policy was undertaken”.
- Define the unavoidable term once. If a technical word must stay, explain it the first time and then use it consistently.
Read the edited passage aloud. A readability formula cannot hear rhythm, but you can. If the shorter version sounds like a machine reading a manual, the original may have been doing more work than the score suggested.
An example
Before
The organization’s implementation of the revised reimbursement procedure requires that employees submit documentation of all expenditures incurred during approved business travel prior to the completion of the reporting period.After
Employees must submit receipts for business travel before the reporting period ends. The revised reimbursement procedure applies to all approved trips. Keep the documentation with your expense report.The original is not wrong; it is doing too many jobs at once. It names the organisation, the procedure, the requirement, the evidence, the timing, and the condition. The revision separates the instruction, the scope, and the record-keeping step. The word count barely changes, but the sentence count rises and the average sentence becomes much shorter — which is exactly the kind of change readability formulas reward.
Notice that the revision also adds a direct actor: “Employees must submit”. The original hides the actor inside “requires that employees submit”. Naming the actor makes the instruction easier to follow even before a formula measures it.
Limits worth knowing
- Syllables are estimated. The counter uses spelling rules, not a dictionary. Proper nouns, abbreviations, and unusual words can be counted incorrectly, which shifts every formula that depends on syllables.
- The formulas are English-specific. They were fitted to English prose. Using the grade bands on another language is not meaningful, even if the syllable counter returns a number.
- Short texts are unstable. A single long sentence can dominate a paragraph. SMOG is particularly rough with fewer than thirty sentences. Use the sentence list instead of treating the score as a verdict.
- Lists, dialogue, and code confuse the sentence splitter. A bulleted list may look like one long sentence. Dialogue may contain fragments. Code is not prose. The analysis is only as good as the boundaries it is given.
- No formula measures understanding. Readability is not comprehension, accuracy, originality, or engagement. A text can score “easy” and still be wrong, and a text can score “difficult” and still be the clearest thing its intended reader has ever read.
Page updated