# LinkedIn Post Readability Checker You wrote something good. Then you read it back on your phone and it felt heavier than it did in the editor, and you cannot say why. Usually it is one sentence. A clause too many in the opening, a word you would never say out loud, an idea that needed a full stop halfway through and did not get one. The rest of the post is fine. Readability formulas are decent at catching that, and they have been around since the 1940s, which is longer than any engagement heuristic anyone will sell you. The catch is that they were built for essays and manuals. A LinkedIn post is neither. This page runs six of them on your draft, shows the arithmetic behind each number, and points at the lines doing the damage. It does not invent a score of its own, and it does not claim any of this changes your reach. ## How it works 1. **Paste your draft**. Drop the post into the box, line breaks and all. We do not send it anywhere. The counting happens in your browser. 2. **Read the six scores**. Flesch Reading Ease is the one conventionally read on a 0 to 100 scale, higher being easier. The other five report a US school grade. Every row will show you the arithmetic that produced it. 3. **Read the two panels separately**. The formulas panel tells you about your vocabulary. The lines panel tells you about your formatting. Neither one answers the other question. 4. **Shorten the longest lines**. The three longest are listed in the lines panel. Splitting the top one usually moves every formula at once. 5. **Reread the flagged words**. The last panel marks the words and phrases that read as machine-written, plus anything that looks passive. None of it touches the scores above. Treat a mark as a reason to read the line once more. ## What you get - Six published readability formulas at once: Flesch Reading Ease, Flesch-Kincaid Grade, Gunning Fog, SMOG, Coleman-Liau and ARI - The arithmetic behind every score, with your own counts substituted into the published formula - Line statistics LinkedIn writers need and no formula measures: non-empty lines, words per line, and the three longest lines - The lines-to-sentences split that shows why other checkers misread a post written in fragments - One Flesch figure recomputed with each line as a unit, labelled an adaptation rather than a Flesch score - Flags for the words, adverbs and phrases that read as machine-written, marked in place in your draft - A possible passive voice check, labelled possible because a be-verb plus a participle is a hint rather than a diagnosis - Runs entirely in your browser. No signup, no upload, and no API call that sees your draft ## Why five short lines confuse most readability tools Readability formulas all divide by sentence count somewhere. Flesch Reading Ease and Flesch-Kincaid use words per sentence directly, Gunning Fog and ARI use it as one of two terms, SMOG divides by it under a square root. So sentence count is not one number among many. Get it wrong and every score moves. Now think about how people write on LinkedIn. Short lines. One idea each. Often no full stop at the end, because the line break already does the work. A formula that only ends a sentence at a full stop reads that as a single sentence of about twenty words. Feed it to Flesch and you get a mediocre score with the note that your sentences are too long. The tool cannot see the line breaks, so it blames the only thing it can see. We tested this. Paste five short unpunctuated lines into aiCarousels, which ranked first for this search when we tested it in September 2026, and it reports one sentence, an average of 22 words per sentence, and advice to try shorter sentences. The longest line in that post is six words. The tool ranking second scores the same post 47 out of 100 on readability and suggests aiming for 10 to 15 words per sentence. Neither is telling you your post is bad, exactly. Both are telling you the problem is sentence length, and the evidence on screen says it is not. Here is what that means for how the numbers are worth reading. Every one of these six formulas takes two kinds of input: a sentence count, and counts of your words, syllables and letters. Line breaks change the first and leave the second untouched, because syllables per word and letters per word never look at a sentence boundary. So the formulas panel on this page is worth trusting on your vocabulary and worth discounting on your structure, and the lines panel next to it covers the structure instead, with counts rather than a score. What we will not do is split on line breaks, print the result, and call it Flesch Reading Ease. Flesch defined a sentence by its full stop. Renaming his formula to suit LinkedIn would make us exactly the kind of tool this page was built to argue with. There is one adapted figure in the lines panel, labelled as an adaptation, and it exists to show you how far a score moves when the segmentation changes. One number is enough to make that point. ## The six formulas, and what each one measures Every score on this page comes with its formula and its inputs, one click away on each row. If you cannot check the arithmetic, you should not trust the number. ### Flesch Reading Ease `206.835 − 1.015 x (words / sentences) − 84.6 x (syllables / words)` Conventionally read on a 0 to 100 scale, higher being easier, though the formula itself is unbounded and very dense or very sparse text will run past either end. We show what it computes rather than clamping it. Rudolf Flesch published the interpretation bands himself: 90 and up is very easy, 60 to 70 is plain English, below 30 is very confusing. It is the only score on this page with bands from the person who invented it, which is why it is the only one that gets a label. ### Flesch-Kincaid Grade `0.39 x (words / sentences) + 11.8 x (syllables / words) − 15.59` Same two inputs, rescaled to a US school grade. It is the common shorthand for "grade level". aiCarousels, which ranked first for this search when we tested it in September 2026, does not offer it at all. ### Gunning Fog `0.4 x [(words / sentences) + 100 x (long words / words)]` Weighs long words much more heavily than the Flesch pair do. If your score here is well above the others, the problem is vocabulary rather than sentence length. ### SMOG `1.0430 x √(long words x (30 / sentences)) + 3.1291` Harry McLaughlin designed it for 30-sentence samples. A LinkedIn post is rarely close to that, so treat it as the loosest number here. We show it anyway, with the caveat, rather than dropping it and leaving you to wonder. It is also the one formula that does not simply add sentence length to word difficulty: the two multiply inside the square root, which is why it can swing harder than the others on short text. ### Coleman-Liau `0.0588 x L − 0.296 x S − 15.8, where L is letters per 100 words and S is sentences per 100 words` No syllables anywhere in it, which makes it the most reliable score on the page, since syllable counting is the part every tool including this one has to estimate. ### ARI `4.71 x (characters / words) + 0.5 x (words / sentences) − 21.43` Also syllable-free, built for typewriters counting keystrokes. Close cousin of Coleman-Liau, with one difference worth knowing: Coleman-Liau counts letters, ARI counts characters, which here means letters plus digits. A post full of numbers will separate the two. Expect the six to disagree, especially on a short post. That is not a bug in any of them. They weight different things and they were calibrated on longer text than yours. ## Words that sound like AI, and where the list came from Readability formulas do not care whether a sentence sounds human. A post can score 75 on Flesch and still read like it fell out of a chat window, because "we leveraged a robust framework to streamline onboarding" is short clauses made of short words. So there is a second pass, unconnected to the six. It runs three word lists and one grammar check over the draft and marks the hits in your own text. The lists came from Oiti's own writer, which is held to them: 26 terms it may not use, plus 9 adverbs and 10 phrases that are usually empty. All three are printed below in full, because a flag you cannot audit is worth nothing. This is opinion rather than research, and saying so is half the point. "Harness" is a real word. "Just" earns its place about half the time. A mark is a reason to read the line once more, not a deduction, and none of it reaches the six numbers above. The passive check is looser still. It looks for a form of "to be" followed by a past participle, so it catches "was written by the team" and it also catches "is excited", which is not passive at all. Hence possible on the label. Participles come from the -ed ending plus a short list of irregular ones. A blanket -en rule got dropped on the way, because it fires on "was open" and "is green", and on a page arguing that other tools publish numbers they cannot justify that is the worse trade. One more limit worth knowing. We match whole words rather than stems, though each single word brings its plural, its past tense and its -ing form along with it, and a multi-word term brings its plural. So "leveraging" and "game changers" both trip the list, and something you welded together yourself out of one of the terms does not. **Words the writer may not use:** delve, foster, leverage, utilize, facilitate, empower, streamline, robust, cutting-edge, paradigm shift, game changer, this is huge, this changes everything, tapestry, realm, beacon, multifaceted, meticulous, intricate, paramount, transformative, elevate, embark, supercharge, harness, ever-evolving. **Adverbs that are usually empty:** just, literally, honestly, simply, actually, truly, fundamentally, importantly, crucially. **Phrases that are usually empty:** it's worth noting, at the end of the day, when it comes to, at its core, in today's world, the reality is, in terms of, in order to, going forward, let's dive in. ## How we count, and where the counting is approximate **Sentences.** A sentence ends at a full stop, question mark or exclamation mark, which is how the formulas define it. Line breaks are treated as ordinary spaces here, so your formatting does not create sentences the formula's author would not have counted. Abbreviations still split: "Mr. Chen" reads as two sentences, and nothing in a browser fixes that cheaply. **Lines.** Counted separately, in the lines panel, and a line is exactly one unit no matter how it is punctuated. Two sentences on one line count as one line and as two sentences, in their respective panels. A line is any run of text between two breaks that has something in it; blank lines are ignored. **Links and email addresses.** Swapped for a single placeholder before anything else runs, so a URL does not end four sentences at its dots. Each one then counts as one word and contributes nothing to the letter, character or long-word counts. That is a choice rather than a law: a 74-character link is not writing, and letting it dominate Coleman-Liau and ARI would tell you less than ignoring it does. **Words.** Whitespace-separated tokens with the surrounding punctuation stripped. Anything with no letter in it is not a word, so a standalone emoji, a bullet glyph and a bare number all count as nothing, though the number still counts as a character for ARI. A hashtag or an @mention counts as one word. Two words welded together by a dash with no spaces around it count as one, because whitespace is the only thing splitting them. **Bold and italic text.** LinkedIn has no rich text, so a bold or italic post is written in the Unicode mathematical alphabets, which is what our own formatter produces. Those are different characters from the letters they imitate, and left alone they would read as words with no letters and no syllables in them. We convert them back to plain letters before counting anything, so a bolded draft and its plain twin score the same. The flagged-words panel reads a bolded draft back to you in plain letters for the same reason. The box and the longest-lines panel keep your own characters. **Syllables.** This is the approximate part, in every tool that computes a Flesch score, including ours. There is no syllable dictionary in your browser, so we count vowel groups and subtract a silent trailing e. It is right most of the time and wrong on compound words, on a lot of proper nouns, and on some words ending in -ed. If your post is full of names and product names, lean on Coleman-Liau and ARI, which do not use syllables at all. **Long words.** Gunning Fog and SMOG both count words of three or more syllables. Fog's original definition excludes proper nouns, familiar compound words, and verbs that only reach three syllables through an -es, -ed or -ing ending. We apply none of those exclusions and count every three-syllable word. Deciding whether a token is a proper noun, or whether an -ing word is a verb, takes judgment a text box does not have, and a rule applied halfway is harder to reason about than one stated plainly. Most implementations skip all three quietly. We would rather tell you. **Short posts.** Below about 100 words the numbers get jumpy. One long word or one missing full stop can swing a score several points, and the tool says so on screen when your draft is under that length. To be clear about where the 100 comes from: it is our own cutoff, not a figure from any of the six papers. We picked a round number in the region where these formulas visibly stop being stable. ## What to change first Split the longest line in half. That moves every score on this page at once, because a sentence count feeds all six of them, and it is usually the only edit a draft needs. After that, look at the words rather than the structure. If Gunning Fog and SMOG sit well above Coleman-Liau and ARI, your vocabulary is the drag, and swapping three long words for short ones will do more than any amount of rearranging. This is the half of the formulas your formatting cannot affect, so it is the half worth taking at face value. Read the first two lines on their own before you post. No formula on this page scores them separately, and none should, because two lines is not enough text for any of these to mean anything. But that is the part someone decides on while scrolling, so read it with your own eyes. Then stop. A readability score measures how much work a sentence asks of a reader. It says nothing about whether the idea was worth reading, and no formula on this page can. If a long sentence earns its length, keep it. ## More free tools - [LinkedIn text post formatter](/tools/linkedin-text-post-formatter) — Bold, italics and line breaks that survive the paste into LinkedIn. - [LinkedIn character counter](/tools/linkedin-character-counter) — Check a draft against the real limit on every LinkedIn field. - [LinkedIn post previewer](/tools/linkedin-post-preview) — See exactly where the feed cuts your post off behind "see more". - [LinkedIn post grader](/tools/linkedin-post-grader) — Grade a draft against five checks and get a letter with the reasons behind it. - [LinkedIn Cringe Translator](/tools/linkedin-translator) — Plain English into LinkedIn speak and back, with every buzzword explained. - [LinkedIn banner maker](/tools/linkedin-banner-maker) — The strip behind your photo, at the size LinkedIn actually renders. ## Frequently asked questions ### What readability score should a LinkedIn post have? There is no published answer to that, and anyone who gives you a precise one made it up. Here is what can be said honestly. Flesch's own bands put 60 to 70 at plain English, and most LinkedIn writing that reads well lands at or above that, because it is written in short lines and everyday words. Treat it as a direction rather than a target. A post that scores badly because one sentence is long is a completely different problem from a post that scores badly all the way through, and no single number can tell the two apart. The list of longest lines can. ### Which of the six should I use? Flesch Reading Ease if you want one number, since it is the only one whose author published bands for it. Coleman-Liau or ARI if your post is stuffed with names, product names or acronyms, since neither of those estimates syllables. Gunning Fog if you suspect the vocabulary rather than the sentence length. And read all of them for vocabulary rather than structure, for the reason above. ### Why do the six disagree? Because they measure different things. Two count syllables, four do not. Two weight long words heavily, two barely notice them. On a 60-word post those differences look enormous even though the text has not changed. Read the spread, not any single number. ### Why show line statistics next to the scores? Because they answer different questions and the scores cannot answer the second one. Every one of these formulas takes a sentence count as one of its inputs. On a post written as short unpunctuated lines, that count is measuring something that is not there, so the score drops and the tool blames your sentences. The lines panel measures your formatting directly, with counts rather than a formula, and leaves the scores to do the job they are good at. Write a normal paragraph with normal punctuation and the formulas panel uses the standard segmentation, so there is nothing for it to miss. The formulas here are the published ones. Where our number differs from another checker's by a point or two, the cause is almost always syllable estimation, which every tool has to approximate. ### Where does the AI-words list come from? From Oiti's own post writer, which is held to all three. The lists are printed on this page in full. They are opinion rather than research and they change none of the six scores. A word on one of them can still be the right word in your sentence. ### Do emojis and hashtags count? An emoji on its own is not a word here. It has no letters and no syllables, so there is nothing for a formula to measure, and an emoji sitting inside a word does not change that word's count either. A hashtag counts as one word with the hash stripped, and so does an @mention. Worth knowing before you stack five of them: a long concatenated hashtag reads as a long word to Gunning Fog and SMOG. ### Is my post sent to a server? No. It is all arithmetic and it all runs in your browser. No API call ever sees your draft, there is no account, and nothing is stored. Close the tab mid-sentence and nothing survives it. ### Will a better score get me more reach? We do not know, and neither does anyone who tells you otherwise. A readability score measures how much work a sentence asks of a reader. Easier reading probably means more people finish the post, and finishing is the precondition for everything else, but that is a reasonable belief rather than a measured finding, and we are not going to put a percentage on it. ### Does this work on text that is not a LinkedIn post? Yes. The formulas are general and the numbers are valid for any English text. The lines panel will happily count a poem or a bulleted list line by line too, which is usually what you want. ## Why we built it The LinkedIn post readability checker is a free tool from Oiti, the AI content engine for LinkedIn. You came here to find out whether a draft reads easily. If the harder question is what to write in the first place, that is what Oiti does: it learns the way you actually talk from what you have already published, and drafts posts that read like you typed them. See [ghostwriting-ai.com](https://www.ghostwriting-ai.com).