The sample essay is 109 words and 625 characters, 9 words over a 100-word limit
A character count changes with what you decide to count. Count every letter, space and line break and you get characters with spaces; strip the spaces and line breaks and you get characters without spaces. Words are counted separately, and bytes measure how much storage the text takes, which matters when a database or an older system sets the limit. The sample text in the counter above, a short scholarship-style essay in 3 paragraphs and 5 sentences, has 625 characters with spaces, 515 without spaces and 109 words. In UTF-8 it takes 629 bytes rather than 625, because the two curly apostrophes in “she’d” and “kids’” take 3 bytes each.
The counter starts with a 100-word limit, so it shows the essay running 9 words over and marks where the overflow begins: at “small, everyday problems” in the last sentence. Press “Select the overflow” and those words are highlighted in the text box, ready to cut or rewrite. Switch the limit to characters, characters without spaces or UTF-8 bytes and every figure updates at once. Below the results, a character mix bar shows how much of the text is letters, digits, spaces, punctuation and emoji, and the sentence chart draws one bar per sentence so you can spot the one that runs on. The sample’s second sentence is 27 words long, the only one above the 25-word guide that GOV.UK uses for clear writing. Read silently at an average adult pace, the essay takes about 27 seconds; read aloud, about 36 seconds.
When you need a word or character count
- College essays, scholarship statements and cover letters: application portals usually state a limit in words or in characters. Pick the same unit in the counter before you start trimming. If a box says “characters” and doesn’t say whether spaces count, plan for characters with spaces. That number is always equal to or larger than the count without spaces, so text that fits with spaces fits either way.
- Forms and databases that limit bytes: a field sized in bytes holds fewer characters once accents, curly quotes or emoji appear. In UTF-8, é and ñ take 2 bytes, curly quotes and the em dash take 3, and an emoji takes at least 4, so a 255-byte field cannot always hold 255 characters. Choose UTF-8 bytes as the limit basis and the overflow marker shows exactly which character tips it over.
- Posts on X: X allows up to 280 characters, but not every character counts the same. Characters up to U+10FF (Latin, Greek, Cyrillic, Arabic, Hebrew, Devanagari, Thai and more) and common punctuation count as 1, while Chinese, Japanese and Korean characters and everything else above that range count as 2, every emoji counts as 2 however it is built, and any link counts as 23 because it is shortened to t.co. A post with one link therefore has 257 left for everything else.
- Text messages: an SMS written in the GSM-7 alphabet holds 160 characters. If the message contains any character outside GSM-7, Twilio sends the whole message in UCS-2, which holds 70 characters per message, and a long message is split into parts of 153 (GSM-7) or 67 (UCS-2) characters. Emoji and the curly quotes (“ ” ’) that word processors insert automatically are not in the basic GSM-7 table, so pasting from a document can quietly cost you more than half the room.
- Meta descriptions: Google’s documentation says there is no limit on how long a meta description can be; the snippet is simply truncated in results as needed to fit the device width. A character count helps you keep the important words near the start, but there is no official number to hit.
- Talks and presentations: adults read aloud at about 183 words per minute on average, according to a review of 77 studies. That puts a 3-minute talk at roughly 550 words and a 5-minute talk at around 900. Paste your script and check the reading-aloud row, then time one real run-through, since speaking pace varies from person to person.
How the counts work
- One character: each character you see on screen counts as 1. The counter uses the browser’s Intl.Segmenter, which follows the grapheme cluster rules in Unicode Standard Annex #29. That means 👍🏽 (thumbs up with a skin tone) and é count as 1 each, even when é has been typed as a plain e plus a combining accent. Some websites count UTF-16 code units instead and get a bigger number; when that happens, the results add a “Count on some sites” row.
- Spaces: ordinary spaces, tabs, line breaks, no-break spaces and the wide ideographic space all count as spaces. Characters without spaces removes all of them. Two extra rows show the count with only spaces removed (line breaks kept) and with only line breaks removed, because some forms count one but not the other. A Windows line break (CR+LF) is treated as one line break.
- Words: a word is a run of text between spaces or line breaks that contains at least one letter or digit. That keeps “don’t”, “state-of-the-art”, “and/or”, “$1,250.50” and “12:30” at 1 word each, and a closed em dash (“word—word”) joins its neighbors into a single word too. A dash, emoji or symbol standing alone between spaces is not a word, so “word — word” is 2 words.
- UTF-8 bytes: RFC 3629 sets the size by code point. Plain English letters, digits, spaces and line breaks (U+0000 to U+007F) take 1 byte; accented Latin letters such as é, ñ and ü take 2; the euro sign, curly quotes, the em dash and Chinese or Japanese characters take 3; and most emoji take 4. An emoji built from several code points, such as one with a skin tone or a family joined with zero-width joiners, takes the sum of its parts.
- Sentences, paragraphs and lines: sentences follow the browser’s version of the Unicode sentence rules: a period, question mark or exclamation mark followed by a space and the next sentence ends a sentence, and so does a line break. A paragraph is a block separated by a blank line. Lines are the number of line breaks plus 1, blank lines included.
- Where the overflow starts: counting from the start in the unit you chose, this is the first character (or the first word, with Words) that takes the total past the limit. With bytes, a 2-byte or 3-byte character that crosses the limit counts as overflow from that character on.
- Reading time: the word count divided by 238 words per minute for silent reading, the average Brysbaert (2019) found for adults reading English non-fiction, and by 183 words per minute for reading aloud. Fiction is read a little faster, at about 260 words per minute.
| Character type | On screen | UTF-8 bytes | UTF-16 units |
|---|---|---|---|
| English letter, digit or space (A) | 1 | 1 | 1 |
| Accented letter (é) | 1 | 2 | 1 |
| Accented letter typed as e + combining accent (é) | 1 | 3 | 2 |
| Euro sign (€) | 1 | 3 | 1 |
| Curly quote (“) | 1 | 3 | 1 |
| Em dash (—) | 1 | 3 | 1 |
| Emoji (😀) | 1 | 4 | 2 |
| Emoji with skin tone (👍🏽) | 1 | 8 | 4 |
| Flag (🇺🇸) | 1 | 8 | 4 |
| Family emoji joined with zero-width joiners (👨👩👧) | 1 | 18 | 8 |
Every row is a single character on screen, yet the storage size ranges from 1 to 18 bytes. Sites that count UTF-16 units see an emoji as 2 or more characters, and X counts every emoji as 2 whatever it is made of.
Same text, different counts
| Text | With spaces | Without spaces | Words | UTF-8 bytes | UTF-16 units |
|---|---|---|---|---|---|
| Hello, world! | 13 | 12 | 2 | 13 | 13 |
| Don’t stop. (curly apostrophe) | 11 | 10 | 2 | 13 | 11 |
| state-of-the-art | 16 | 16 | 1 | 16 | 16 |
| word—word (closed em dash) | 9 | 9 | 1 | 11 | 9 |
| word — word (spaced em dash) | 11 | 9 | 2 | 13 | 11 |
| Meet me at 12:30. | 17 | 14 | 4 | 17 | 17 |
| $1,250.50 | 9 | 9 | 1 | 9 | 9 |
| Line one (line break) Line two | 17 | 14 | 4 | 17 | 17 |
| 👍🏽 Nice work | 11 | 9 | 2 | 18 | 14 |
| café | 4 | 4 | 1 | 5 | 4 |
| café (e + combining accent) | 4 | 4 | 1 | 6 | 5 |
A line break counts as 1 character with spaces and 1 byte in UTF-8. The two “café” rows look identical, but the second was typed with a separate combining accent, as can happen when text is copied from file names on a Mac. On screen both are 4 characters; in UTF-16 the second is 5. X normalizes text to the precomposed form before counting, so it sees both as 4.
How to cut down to a limit
- Start where the overflow starts: the preview under the progress bar shows the words just before and after the cut-off point. If only the last sentence spills over, tightening that sentence is often enough, and you keep the rest of the essay intact.
- Split or shorten long sentences: in the sentence chart, dark bars are sentences longer than 25 words, the point at which GOV.UK’s writing guidance suggests splitting a sentence. Click a bar to select that sentence in the text box. A sentence carrying two ideas usually splits cleanly into two shorter ones, and often one of them can go.
- Cut repeated and filler words: the “Most repeated words” list shows words used twice or more, leaving out very common words such as “the”. A word that keeps coming back can often be replaced by a pronoun or merged. Phrases such as “in order to”, “the fact that” and “very” are usually the cheapest words to lose.
- Keep the specifics: details such as “40 to 60 people every Saturday” or “from 18% to 6%” in the sample essay are short and persuasive. Trim general statements and intensifiers first, and keep the numbers and concrete examples that only you can provide.
- Check the counter you will submit to: some tools split words at hyphens, slashes and dashes as well as at spaces, so “state-of-the-art” counts as 4 words there and your total can come out higher than here. Paste the final text into the real form once before you submit.
Common mistakes
- Comparing against the wrong unit: a 650-word limit and a 650-character limit are very different. In the sample essay, 625 characters hold 109 words, about 5.7 characters per word including spaces, so 650 characters would be only about 110 words of similar prose. Check the unit in the instructions, then set the same unit in the counter.
- Assuming every word counter agrees: this counter splits words only at spaces, but some tools follow the Unicode default word rules, which also split at hyphens, slashes and dashes. There “state-of-the-art” is 4 words, “and/or” is 2 and “12:30” is 2, so a text full of compounds and times can differ by several words between tools.
- Forgetting line breaks: some systems store each line break as two characters (CR and LF), so a text with many paragraphs can come out a few characters longer after you submit it. If you are right at a character limit, leave a little room for each line break.
- Treating X’s 280 as a plain character count: emoji count as 2 and every link counts as 23, so a post with a link and a few emoji holds noticeably less text than 280 characters with spaces would suggest.
- Pasting curly quotes and emoji into text messages: a single character outside GSM-7 switches the whole message to UCS-2, cutting the room per message from 160 characters to 70. Straight quotes (" and ') are in the basic GSM-7 table; curly ones are not.
- Mixing up characters and bytes: “255 characters” and “255 bytes” are the same only for plain English text. Every accented letter, curly quote or emoji uses extra bytes, so check the UTF-8 bytes row when the limit is stated in bytes.
Limits and caveats
- Many sites don’t publish exactly how they count. This counter shows the widely used measures side by side but cannot guarantee that any particular site will show the same number.
- Words are runs of text between spaces that contain a letter or digit. Hyphenated compounds, “and/or” and times such as 12:30 count as 1 word each, and a closed em dash joins the words on either side into 1; a dash, emoji or symbol standing alone between spaces is not counted.
- Sentence detection follows the browser’s sentence rules, which do not know abbreviations. A title such as “Dr.” or “Mr.” followed by a capitalized name is counted as the end of a sentence, and a line without final punctuation, such as a heading, counts as a sentence of its own.
- Reading times are averages: Brysbaert’s review found that most adults read English non-fiction silently at 175 to 300 words per minute. Difficult material, unfamiliar vocabulary and reading on a small screen all slow people down.
- The repeated-words list counts words exactly as written, without reducing them to a root form, so “book” and “books” are listed separately. Very common words and words of a single letter are skipped.
- Your text is counted inside this browser and is never sent to a server or saved. The page address only keeps the limit and the limit basis.
Sources
- 01Unicode Standard Annex #29 — Unicode Text SegmentationBoundary rules for user-perceived characters (grapheme clusters) and sentences used by this counter; its default word rules split at hyphens, which is why some other tools count more words (revision 49, 2026-09-01, checked 2026-10-11)
- 02MDN — Intl.SegmenterThe browser feature this counter uses to split text into graphemes, words and sentences
- 03RFC 3629 — UTF-8, a transformation format of ISO 106461 to 4 bytes per code point, by code point range
- 04Brysbaert M. How many words do we read per minute? A review and meta-analysis of reading rate. Journal of Memory and Language 109:104047, 2019Silent reading by adults in English: 238 words per minute for non-fiction, 260 for fiction, most adults between 175 and 300 for non-fiction; reading aloud: 183 words per minute (77 studies, 5,965 participants). Abstract checked 2026-10-11 through the PsyArXiv preprint (doi:10.31234/osf.io/xynwg)
- 05GOV.UK — Use clear language“Try to split up sentences that are over 25 words long.” Checked 2026-10-11
- 06X Developer Platform — Counting charactersUp to 280 characters; Latin, punctuation and common symbols weigh 1, CJK and other Unicode 2 (the twitter-text v3 configuration gives weight 1 to U+0000–U+10FF and the U+2000–U+200D, U+2010–U+201F and U+2032–U+2037 punctuation ranges); every emoji 2; every URL 23; text normalized to NFC before counting. Checked 2026-10-11
- 07Twilio — What is GSM-7 character encoding?160 characters per SMS in GSM-7; the basic table includes straight quotes but not curly ones. Checked 2026-10-11
- 08Twilio — What is UCS-2 character encoding?Any non-GSM-7 character switches the message to UCS-2, 70 characters per message; long messages split into 153 (GSM-7) or 67 (UCS-2) characters. Checked 2026-10-11
- 09Google Search Central — Control your snippets in search results“There’s no limit on how long a meta description can be”; snippets are truncated as needed to fit the device width (page updated 2026-04-20, checked 2026-10-11)
Frequently asked questions
Do spaces count as characters?
It depends on the form, so follow its instructions first. Characters with spaces counts every space and line break; characters without spaces removes them. If the instructions don’t say, plan for the count with spaces: it is always equal to or larger than the count without spaces, so text that fits with spaces fits either way. Some sites remove spaces but still count line breaks, which is the “Spaces removed, line breaks kept” row.
Is my text saved or sent anywhere?
No. Characters, words, bytes, sentence lengths and repeated words are all calculated inside your browser. The text is not sent to a server or stored, and the page address only keeps the limit and the limit basis, never the text.
How many words is 1,000 characters?
It depends on how long the words are. In the sample essay, 625 characters with spaces hold 109 words, about 5.7 characters per word, so 1,000 characters of similar prose is roughly 175 words. Technical writing with longer words fits fewer. Paste your own text to see both counts side by side.
How does X count characters in a post?
X allows up to 280 characters but weights them. Characters up to U+10FF (Latin, Greek, Cyrillic, Arabic, Hebrew and others) and common punctuation count as 1, Chinese, Japanese and Korean characters and the rest of Unicode count as 2, every emoji counts as 2 however it is built, and every link counts as 23 whatever its length. Text is normalized to NFC before counting (X developer documentation, checked October 11, 2026). A post with one link has 257 left for everything else.
How are emoji counted?
This counter counts each emoji you see as 1 character, including emoji built from several code points such as 👍🏽 or a family emoji. Sites that count UTF-16 units see 2 or more (8 for 👨👩👧), and X counts every emoji as 2. In UTF-8 a simple emoji takes 4 bytes and combined ones take more: 👍🏽 is 8 bytes and 👨👩👧 is 18.
How many words should a 3-minute speech be?
Adults read aloud at about 183 words per minute on average, according to a review of 77 studies (Brysbaert, 2019), so 3 minutes is roughly 550 words and 5 minutes roughly 900. Speaking pace varies a lot between people, so paste your script, check the reading-aloud time and then time one real run-through.
Why does another tool give a different word count?
Tools draw word boundaries differently. This counter counts runs of text between spaces that contain a letter or digit, so “state-of-the-art”, “and/or” and “12:30” are 1 word each, and a dash or emoji standing alone is not counted. Tools that follow the Unicode default word rules also split at hyphens, slashes and the colon in a time, giving 4, 2 and 2. For a submission, the form’s own counter is the one that matters.
How many characters fit in a text message?
A single SMS holds 160 characters in the GSM-7 alphabet. If the message contains any character outside GSM-7, Twilio sends the whole message in UCS-2, which holds 70, and longer messages are split into parts of 153 (GSM-7) or 67 (UCS-2) characters. Emoji and curly quotes (“ ” ’) are not in the basic GSM-7 table, so swap curly quotes for straight ones when you are close to the limit.