Word & Character Counter
Words, characters, sentences and reading time — counted as you type, never uploaded.
238 is the measured average for silent reading of English prose.
Counted entirely in this page. Nothing you paste is transmitted, stored, or seen by anyone — which matters when the text is a draft contract, a cover letter or an unpublished manuscript.
How it works
Counting words sounds trivial until you ask what a word is. "Don't" is one. "Mother-in-law" is one to most people and three to a naive splitter. "$1,500" is one. An em-dash between two words — like that — is not a word. This splits on whitespace and then discards anything with no letter or digit in it, which is the rule that matches how a person counts and how word processors count.
The number people usually actually want is characters, and there the trap is what counts as one. Emoji are frequently two UTF-16 code units, so a naive `length` in JavaScript reports 2 for a single 😀 and gets worse for flags and skin-tone variants, which can be seven or more. This counts by grapheme — what a human sees as one character — and separately reports UTF-16 length and UTF-8 bytes, because those are what a database column limit and an SMS gateway actually measure.
Reading time uses 238 words per minute, which is the figure from the largest meta-analysis of silent reading rates for English prose rather than the 200 or 250 that gets quoted without a source. Speaking is much slower: 130–150 is a comfortable presentation pace, and that is what the estimate here uses.
Keyword density is included because it is asked for constantly, and it is worth saying plainly that it has not been a ranking factor for many years. It is useful for spotting that you have written "solution" nine times in four paragraphs. It is not a dial to tune.
Common questions
Does it count the same way as Word or Google Docs?
For ordinary prose, yes. All three split on whitespace and ignore standalone punctuation. Differences show up with hyphenated compounds and with numbers containing separators, where processors disagree with each other too.
Why do my emoji count as more than one character?
They do not, in the grapheme count — that is the number matching what you see. The other two figures are there because they are what systems enforce: UTF-16 length is what most databases and JavaScript measure, and UTF-8 bytes is what an API payload limit counts.
What reading speed should I use?
238 words per minute is the average for silent reading of English prose, from a meta-analysis of nearly two hundred studies. Technical material runs slower, around 150–200. Reading aloud is slower still, near 130–150, which is the speaking estimate here.
Is keyword density worth optimising?
No. It stopped being a ranking signal a long time ago and deliberately hitting a percentage produces text that reads badly. It is useful for noticing repetition you did not intend, and nothing more.