Guides › Text & developer

Why does one emoji count as 2 characters? Character limits, emoji and SMS, tested

You paste a bio into a form with a 150-character limit and it is rejected, even though you count fewer. Or a text message that should fit in one SMS arrives as three. Both come from the same fact: "character" means different things to different software. We counted the same eight texts four ways, using our word and character counter and the standard methods behind most limits.

Short answer: most websites count UTF-16 code units, the unit JavaScript uses. Ordinary letters count 1, but most emoji count 2 and combined emoji such as families count up to 8 or more. For SMS, a single character outside the basic GSM alphabet, such as a curly apostrophe (’) or an emoji, switches the whole message to a format that holds 70 characters instead of 160.

The measurements

TextSymbols you seeOur counter / most websitesBytes (UTF-8)SMS encoding
Meeting moved to 3pm, room 204. Thanks!!404040GSM-7, 40 of 160
Can’t make it today, see you tomorrow ok404042UCS-2, 40 of 70
Total: €25 for two tickets, pay at door.404042GSM-7, 41 of 160
café445GSM-7, 4 of 160
안녕하세요 (Korean "hello")5515UCS-2, 5 of 70
👍124UCS-2, 2 of 70
🇰🇷 (flag)148UCS-2, 4 of 70
👨‍👩‍👧 (family)1818UCS-2, 8 of 70

"Symbols you see" are grapheme clusters, the unit defined by the Unicode standard for what a reader perceives as one character, which we counted with the browser's Intl.Segmenter. "Our counter" is the value our tool shows as characters, which matches JavaScript's string length and the maxlength limit on standard web form fields.

Why emoji count as more than one

Computers store text as numbered code points. Common letters, including Korean and accented letters, fit into a single 16-bit unit in the UTF-16 encoding that JavaScript, Java and Windows use. Most emoji were added later at code points beyond that range and need two units each.

Many emoji are also built from several code points. A flag is two "regional indicator" letters (K + R for Korea), so 4 units. The family emoji is man, woman and girl joined by invisible "zero-width joiner" characters: 5 code points, 8 units. Skin-tone and gender variations add more. Your phone shows one picture; a character limit sees the parts.

How SMS counts

A standard SMS carries 140 bytes. Phones fit 160 characters into that by using the GSM 7-bit alphabet, which contains basic Latin letters, digits, common punctuation and some accented letters such as é and ü. A few symbols, including € [ ] { } and ~, take two slots, which is why our euro example used 41 of 160.

If the message contains any character outside that alphabet, the phone switches the whole message to UCS-2, a 16-bit encoding, and the limit drops to 70. Common triggers are curly quotes and apostrophes that phones and word processors insert automatically, emoji, and non-Latin scripts such as Korean, Chinese or Arabic.

EncodingSingle messagePer part when splitOur test: parts needed
GSM-7160153160 chars → 1; 161 → 2; 307 → 3
UCS-2706770 → 1; 71 → 2; 135 → 3

Long messages are split into parts that each lose a few characters to a header used to reassemble them. Business messaging services usually charge per part, so a single curly apostrophe in a 100-character marketing text can double its cost.

How major limits count

Practical tips

  1. When writing to a limit, paste the text into the character counter and treat its number as the strict count; if the site uses visible symbols instead, you will have room to spare.
  2. For SMS, replace curly quotes and apostrophes with straight ones (' and ") and avoid emoji when you need the 160-character format.

Fifty emoji, counted

To see how often an emoji costs more than 2, we typed 50 emoji into our counter one at a time. We chose them ourselves: mostly common faces, hands and hearts, plus a skin-tone variant, three flags and several combined emoji, so the split below describes this set rather than all emoji.

We also typed messages of exactly 160, 161, 306 and 307 plain letters, and 70, 71, 134 and 135 Korean syllables, and the counter showed exactly those numbers; the SMS parts for each length, worked out from the GSM-7 and UCS-2 rules, match the table above.

You can download the data and check it yourself. The .csv files open in any spreadsheet and start with a header row; they are saved as UTF-8 so the emoji display correctly.

Common questions

Does a space count as a character?

Yes, in almost all limits. Our counter shows the count with and without spaces.

Why does my phone say a message is "2/3" when it is short?

It probably contains an emoji or curly quote, which switched the message to the 70-character format.

Are Korean characters counted as 2 or 3?

One in UTF-16 (our counter and most web forms), two on some social networks, and three bytes in UTF-8 storage.

References

  1. Unicode Standard Annex #29: Unicode Text Segmentation (grapheme clusters)
  2. 3GPP TS 23.038: Alphabets and language-specific information (GSM 7-bit default alphabet)
  3. MDN: String length (UTF-16 code units)
  4. MDN: Intl.Segmenter

Written and checked by Ria Kim for GlowHalo; figures without a source link come from the steps shown on this page. Companion tool: Word Counter. Spotted an error? Write to help@glowhalo.org. Windows is a trademark of its owner; GlowHalo is not affiliated with it.

More guides

Text & developerUUID v4 or v7 as a database primary key? We inserted a million rows of eachRandom v4 keys were 3–4.7 times slower to insert than time-ordered v7 keys in our SQLite test. What causes the gap and when it matters.Read guide →Text & developerIs your random picker really fair? Modulo bias, measured on 6.2 million drawsA common shortcut made eight characters 25% more likely than the rest in our test. How fair random picks are made, and how to check a tool.Read guide →Text & developerHow big should a printed QR code be? What data length and error correction do to itWe generated QR codes for five real payloads at all four error-correction levels and counted the modules. Shorter data matters more than anything else.Read guide →