Word & Character Counter

Counting is less obvious than it looks

Different tools give different numbers for the same text, and none of them is wrong exactly; they are applying different rules. Does a hyphenated compound count as one word or two? Is an em dash between words a separator? Does a standalone numeral count? Does a URL count as one word or as its parts? This tool splits on whitespace after normalising runs of spaces, which matches the most common convention and what most word processors report.

Characters with and without spaces

Both figures matter in different places. Publishers and translators usually bill on characters including spaces. Platform limits are usually characters excluding nothing at all: a tweet, an SMS segment, a meta description. Note that character counts and byte counts diverge for non-ASCII text, since a single emoji can occupy four bytes and some emoji are sequences of several code points joined by zero-width joiners, which is why an emoji can consume far more of a limit than it appears to.

Reading time is an estimate, not a measurement

The usual figure is 200 to 250 words per minute for adult silent reading of general prose. That average hides enormous variation: technical documentation with code samples reads far slower, and skimming is far faster. Treat a reading-time label as a rough expectation-setting device rather than a fact, and be aware that showing one changes behaviour, since readers use it to decide whether to start at all.

Where limits actually bind

A meta description is truncated around 155 to 160 characters. A page title is truncated around 60. SMS is 160 characters for a single GSM-encoded segment, dropping to 70 if any character forces UCS-2 encoding, which one emoji or one curly apostrophe will do. Academic abstracts are commonly capped at 250 or 300 words. Knowing which unit a limit uses matters as much as the number.

Local processing

Counting happens in your browser, so drafts, submissions and client work are never transmitted.