Flesch Readability Checker
Score text with Flesch Reading Ease so captions and landing pages stay easy to scan on a phone.
- Flesch score
- -
- Reading level
- -
- Words
- 0
- Sentences
- 0
Flesch Reading Ease scores 60-70 are ideal for web content. Higher = easier to read. Scores are estimates - edit for clarity regardless of the number.
Runs in your browser. Nothing you enter is uploaded.
Overview
Flesch Reading Ease turns two measurable properties of your writing - how long your sentences run and how many syllables your words carry - into a single number between roughly 0 and 100. Higher means easier. This checker computes that score in the browser as you paste, alongside the raw word and sentence counts it used.
The score is not a verdict on quality. A brilliant paragraph full of technical nouns will score low, and a run of choppy filler sentences will score high. What the number is genuinely good at is catching drift: the moment a draft slides into 40-word sentences and Latinate vocabulary without you noticing.
Use it on a finished draft rather than an outline. Outlines are fragments, and fragments confuse the sentence detection that the formula depends on.
What a Flesch score actually rewards
The formula has exactly two levers. Cutting average sentence length raises the score. Choosing words with fewer syllables raises it more sharply, because the syllable term carries the heavier coefficient. Nothing else - not paragraph structure, not logical order, not whether the argument makes sense - enters the calculation.
That narrowness is why the score can be gamed and why gaming it is pointless. Splitting every sentence at the comma will lift the number and make the prose worse. The productive reading is directional: if a page you rewrote for clarity scores twelve points higher than the original, the rewrite probably did shorten the reader’s work.
Choosing a target band for the page you are writing
General web copy, landing pages, and how-to articles sit comfortably in the 60-70 range, which the tool labels Standard. Consumer newsletters and social-adjacent writing often land in the 70s. Documentation for practitioners, legal explanations, and medical or financial content routinely score in the 30s and 40s because the required vocabulary is polysyllabic, and forcing those pages upward usually means removing precision.
Set the band by audience rather than by ambition. If the reader is a specialist searching for an exact term, the term stays even though it costs you points. If the reader is a first-timer who typed a plain-language question into search, the specialist term needs a plain-language sentence next to it.
- 90-100 (Very Easy): short declaratives, common words - suits children’s content and simple instructions.
- 60-70 (Standard): the default target for most marketing and how-to pages.
- 30-50 (Difficult to Fairly Difficult): normal for technical, legal, and academic subject matter.
- Below 30 (Very Difficult): worth a second look even in technical writing - usually a sentence-length problem, not a vocabulary one.
The formula and the shortcuts this implementation takes
The score is the standard Flesch Reading Ease equation: 206.835 minus 1.015 times the average words per sentence, minus 84.6 times the average syllables per word, rounded to a whole number. Words are whitespace-separated tokens. Sentences are counted by splitting on periods, question marks, and exclamation points and discarding empty fragments.
Syllables are estimated, not looked up in a dictionary. The heuristic counts groups of vowels including y, treats any word of three letters or fewer as one syllable, and subtracts one for a silent trailing e unless the word ends in "le". This is fast and approximately right across ordinary English, and it is predictably wrong on a few shapes: it undercounts words like "area" and "poem" where adjacent vowels form separate syllables, and it can overcount unusual proper nouns.
Because the estimate errs in both directions, the error largely washes out over a few hundred words. Over a single sentence it does not, which is why the tool asks for at least 100 words before the number means much.
One paragraph, before and after editing
Input
In order to facilitate the optimisation of your content distribution strategy, it is recommended that practitioners undertake a comprehensive evaluation of historical performance data prior to the implementation of any substantive modifications to their existing publication cadence.
Output
Flesch score: 3 · Reading level: Very Difficult · Words: 38 · Sentences: 1
One 38-word sentence built almost entirely from four- and five-syllable nouns produces a near-floor score. Rewritten as "Before you change how often you publish, look at what your past posts actually did," the same instruction scores in the high 80s at 15 words. Nothing was lost except the abstraction - which is the point the number was flagging.
Limitations
- Syllable counting is heuristic. Words with adjacent vowels in separate syllables (area, video, poem) and irregular proper nouns are the reliable failure cases.
- Sentence detection splits on . ! and ? only, so abbreviations like "e.g." and decimal numbers inflate the sentence count and push the score up artificially.
- The formula was derived for English. Running German, Spanish, or any non-English text through it produces a number, but not a meaningful one.
- Lists, headings, and code blocks are treated as prose. Pasting a page that is mostly bullet fragments will report a misleadingly easy score.
- Readability is not a documented Google ranking factor. A high score does not lift rankings on its own.
When not to use this tool
- For short copy where the constraint is a character cap rather than comprehension - bios, ad headlines, meta descriptions - a character counter answers the actual question.
- For deciding article length or reading time, the reading time calculator is the direct tool; Flesch says nothing about how long a page takes to read.
How to use it
- 1
Paste at least 100 words of finished body copy - not an outline - into the text area.
- 2
Read the Flesch score together with the sentence count; a low score with few long sentences points at vocabulary, a low score with many words per sentence points at structure.
- 3
Split the longest sentences first, since sentence length is the lever you can pull without changing meaning.
- 4
Re-paste the edited draft and compare the two scores rather than chasing a fixed target number.
Questions
My draft scores 45 and I cannot get it higher without losing accuracy. Is that a problem?
Not if the subject demands the vocabulary. A page about schema markup has to say "structured data" and "JSON-LD", and both cost syllables. Check the sentence-length half of the equation instead: if your average sentence is under about 20 words and the score is still low, the vocabulary is doing the work and the score is telling you about the topic, not the writing.
Why did my score jump when I added subheadings?
Headings usually lack terminal punctuation, so they get absorbed into the following sentence or counted oddly by the split. Score the body copy on its own if you want a stable number across drafts, and keep the same approach every time so the comparison stays valid.
Does the word count here match what my CMS reports?
It counts whitespace-separated tokens, so hyphenated compounds count as one word and an em dash surrounded by spaces counts as one. Most CMS counters use the same rule, but some strip punctuation first. Small differences are normal; large ones usually mean one of the two is including or excluding markup.