What Reading Level Should Your Essay Be?

Somebody has told you your writing is too dense. Or a supervisor asked for "grade 12", or a journal mentioned a target Reading Ease, or you pasted a paragraph into a checker and got a number you had no way to interpret. So: what reading level should an essay be?

The honest answer is that the question has a good version and a bad version, and most advice online answers the bad one.

The bad version is "what number should I aim for". The good version is "what is this number telling me about my sentences, and which part of that is worth fixing". This post is about the second one, because chasing the first is how people end up with writing that scores well and reads worse.

What the number actually measures

Every readability formula in common use — Flesch Reading Ease, Flesch-Kincaid, Gunning Fog, SMOG, Coleman-Liau, the Automated Readability Index — is arithmetic over three counts: how many words, how many sentences, and how long the words are, measured in either syllables or letters.

That is the whole of it. None of them reads your text. None can tell whether your argument is ordered sensibly, whether you defined a term before using it, or whether the paragraph you are proudest of actually makes sense.

What they do capture is real, though, and worth taking seriously: mechanical load. A long sentence makes a reader hold more in working memory before the clause resolves. A long word is more often the abstract, technical or Latinate one. Both cost the reader effort, and both are measurable without understanding a single sentence.

So the number is a proxy. It is a decent proxy for "how hard is this to process", and a terrible proxy for "how good is this".

The rough answer, with the caveats attached

For undergraduate coursework in the humanities or social sciences, writing that lands around grade 12 to 15 on Flesch-Kincaid, or 30 to 50 on Flesch Reading Ease, is entirely normal. Graduate work and journal articles routinely sit higher — grade 16 to 18, Reading Ease in the 20s — and that is not a defect.

If you are writing for a general audience, a press release, a patient information leaflet or a public-facing report, the targets are genuinely lower: grade 8 to 10, Reading Ease 60 or above. Health communication in particular has a well-established standard around grade 6 to 8, which is where SMOG gets used.

Now the caveat that matters more than the numbers. None of these formulas was validated on academic prose. Flesch-Kincaid was fitted in 1975 on US Navy training manuals for enlisted personnel. Gunning Fog came out of business writing in 1952. SMOG was built for health materials in 1969. Coleman-Liau and the ARI were designed for machine scoring, which is why they avoid syllables entirely.

Applied to a paper about, say, mitochondrial biogenesis, every one of them reads harder than the audience finds it — because a specialist reader knows "mitochondrial" perfectly well and the formula only knows it has six syllables. The absolute number is approximate. The change between your drafts is the reliable signal.

Lower is not better

This is the part that gets lost. A readability score can always be improved by removing precision, and removing precision is usually the wrong trade.

"We administered the intervention to participants" scores worse than "We gave it to them". The second is easier to read and says less. If your discipline's readers need to know which intervention and which participants, the harder sentence is the better one.

The rule of thumb worth holding: shorten sentences before you simplify words. Sentence length is the input every formula shares, it is usually where the load actually is, and it costs you nothing. A forty-word sentence carrying three ideas becomes three sentences carrying one each, with no vocabulary lost and no precision surrendered.

Only after that is done is it worth looking at the long words — and then only at the ones doing no work. "Utilise" for "use". "Necessitate" for "need". "In order to" for "to". "The majority of" for "most". Those cost you readability and buy nothing. The technical terms you actually need are a different matter, and you should keep them.

Why the formulas disagree, and what the gap tells you

Run the same paragraph through six formulas and you will get six numbers, sometimes several grades apart. That looks like a flaw. It is the most useful thing they do.

The six split into two families:

Syllable-based — Flesch Reading Ease, Flesch-Kincaid, Gunning Fog, SMOG. These punish long words hard.
Letter-based — Coleman-Liau, ARI. These count letters per word instead, which sidesteps syllable-counting error entirely.

So when Gunning Fog reads several years above Coleman-Liau, your problem is vocabulary: a lot of polysyllabic words relative to their letter count. When both are high and close together, your problem is sentence length. That diagnosis is far more actionable than any single number, and you only get it by looking at more than one formula.

There is a second reason to distrust a single score. Syllable counting in English is a heuristic, because English spelling does not reliably encode syllables. "Business" and "poem" are both commonly miscounted. Any formula built on syllables inherits that error, which is precisely why Coleman-Liau and the ARI were designed to avoid them.

Our Readability Checker runs all six at once and shows what each is measuring, along with the counts behind them, so you can see which family is driving the result rather than taking one number on faith. It also flags when a formula is being used outside its own stated limits — SMOG, for instance, was defined on thirty-sentence samples and is not reliable on a single paragraph.

What to actually do with the score

A workable process, in order:

Run the whole document, not a paragraph. Ratios over short text are noise; one long sentence moves a grade by years.
Look at the gap between the syllable and letter formulas. That tells you whether to attack sentences or words.
Find the two worst paragraphs. Not every paragraph — the two that are dragging the average. Fix those.
Re-run and compare to your own earlier draft, not to a target. The delta is trustworthy in a way the absolute number is not.
Stop. There is a point past which you are optimising for a 1975 formula rather than for a reader.

If a supervisor has genuinely asked for a specific grade level, give them that number — but understand that what they are asking for is prose a competent reader can follow, not a specific output from a fifty-year-old regression.

The related problem a readability score cannot see

Readability formulas are blind to repetition. A paragraph that uses "however" five times, opens four consecutive sentences the same way, and leans on the same verb throughout can score beautifully — short sentences, common words — while being genuinely tiring to read.

That is a different measurement, and if your writing feels monotonous rather than dense, the Lexical Diversity checker is looking at the right thing: vocabulary variety, overused words, repeated sentence openers, and runs of sentences that are all one length. We wrote about that in more detail in why your writing sounds repetitive.

Between them the two cover most of what "this is hard to read" turns out to mean in practice. Neither can tell you whether the argument is any good. That part is still yours.