Character Counter

Why do two character counters give different numbers?

Because there are four different things that can reasonably be called a character, and different tools count different ones. A grapheme cluster is what a person sees as one character. A code point is one Unicode value. A UTF-16 code unit is what JavaScript's String.length returns, which is what many counters report without saying so. A UTF-8 byte is storage size, which is what a database column limit usually means.

For ordinary English text all four agree, so the disagreement never shows up. Add one emoji and they split apart immediately. A thumbs-up with a medium skin tone is a single grapheme, but it is built from two code points, occupies four UTF-16 code units, and takes eight UTF-8 bytes. A four-person family emoji is still one grapheme, yet it is seven code points, eleven UTF-16 code units and twenty-five UTF-8 bytes.

Accents can split the same way. The letter e-acute can be stored as one precomposed code point or as a plain e followed by a combining accent. Both look identical on screen and both are one grapheme, but the second version is two code points and three UTF-8 bytes instead of one and two. This is why X normalizes text to Unicode Normalization Form C before counting anything, so the two spellings cost the same.

The counter above shows all four numbers together, so when a platform rejects text that your counter said was short enough, you can see which measure it was actually using.

Why is the SMS character limit 160, or 70 with an emoji?

An SMS carries a fixed payload of 1,120 bits. The original GSM 03.38 alphabet packs each character into 7 bits, and 1,120 divided by 7 is 160, which is where the famous limit comes from. That alphabet covers basic Latin letters, digits, common punctuation and a short list of European characters. Nine more characters, including the square brackets, curly braces, backslash, caret, tilde, pipe and euro sign, live in an extension table and cost two septets each.

The moment your message contains a single character outside that alphabet, such as an emoji, a curly apostrophe pasted in from a word processor, or any non-Latin script, the whole message switches to 16-bit UCS-2 encoding. At 16 bits per character the same 1,120-bit payload holds only 70 characters. One smart quote can therefore cut your limit by more than half.

Longer messages are not truncated, they are split into segments and reassembled on the handset. Each segment gives up seven bits of space to a User Data Header describing how to rejoin them, which drops the capacity to 153 characters per segment in GSM-7 and 67 in UCS-2. Carriers bill per segment, so a 161-character message costs twice what a 160-character one does. The counter above shows the encoding, the segment count, and which character forced the switch when one did.

How many characters should a meta title or description be?

Neither has a published limit, which is the honest answer that most character-count content skips. Google's own documentation states that there is no limit on how long a meta description can be, and that the snippet is simply truncated in search results as needed, typically to fit the device width. The same is true of the title link, which Google may also rewrite entirely.

Truncation happens on pixel width, not on character count, so a title made of wide capitals clips sooner than the same number of narrow lowercase letters. The commonly quoted figures of roughly 50 to 60 characters for a title and around 155 for a description are useful working targets derived from how much fits a typical desktop result, but they are conventions rather than rules. Front-load the important words and the exact count matters much less.

What do Cut here and Copy trimmed to fit do?

When your text runs over a limit, the row for that field shows Cut here, followed by the last few words before the point where the text has to end, so you can see where the cut falls without counting. Copy trimmed to fit then copies everything up to that point. The text in the editor is not changed; only the copy is shortened.

The cut is measured in the unit that field counts in: X weighted units for an X post, so an emoji costs 2 and a link 23; graphemes for a Bluesky post; characters for a meta description, a title tag and the other fields; and segments for an SMS, so the trimmed copy is sent as a single message. Under Your limit it is measured in characters or words, whichever you picked.

The cut moves back to the end of the last whole word, so no word is split in half, and it never falls inside an emoji, a flag or a letter with its accent. When the first word alone is already over the limit, a long link for example, the cut falls inside that word instead, so there is always something to copy, and the row says so.

Common lengths: which fields use them and how many words fit

Each length is matched against the 15 platform fields in WordBench's limits table, verified on 23 July 2026, with the unit a field counts in when it is not plain characters. Words are the length divided by 5, the typing-test convention, and by 5.7, the average English word of 4.7 characters plus the space after it, rounded to the nearest whole word.

A limit counted in weighted units or graphemes is not the same as a plain character count: on X an emoji costs 2 and a link always costs 23. Short, common words fit more words into a length than long or technical ones, so the word figures are estimates, not limits.

How the same text measures in four different units

Every row is a single user-perceived character. The columns are the four ways a tool might count it.

Counts follow the Unicode encoding rules: UTF-8 uses one to four bytes per code point, and any code point above U+FFFF takes two UTF-16 code units as a surrogate pair.