Words, characters, sentences and reading time as you type. Nothing is uploaded — the counting happens in your browser.
0
Words
0
Characters
0
No spaces
0
Sentences
0
Paragraphs
0
Lines
0
Reading
0
Speaking
Most used words
Start typing and the statistics appear as you go.
01 · what it counts
Most limits you are writing to are character limits, and they do not agree about spaces. A tweet counts them; some database fields do not; a form that says "500 characters" almost never says which it means. Showing both counts means you can check against whichever definition the thing in front of you is using.
The word count follows the ordinary convention: a run of letters or digits, joined by an apostrophe or a hyphen where there is one. "Don't" is one word, "twenty-one" is one word, and a number counts as a word. That is close enough to what a word processor does that the two rarely differ by more than a word or two on a page.
Sentences and paragraphs are approximations and honest about it. A sentence ends at a full stop, question mark or exclamation mark, which over-counts slightly on text full of abbreviations and decimals.
02 · the limits
Nobody counts words for fun. These are the limits that send people to a counter, and each one has a wrinkle worth knowing.
| Where | Limit | The wrinkle |
|---|---|---|
| Meta description | 155 characters | Google truncates around there on desktop. The limit is really pixel width, so a line of capitals is cut sooner. |
| Meta title | 60 characters | Same reasoning. Front-load the words that matter; the tail is the part that disappears. |
| SMS | 160 characters | A single message. Past it the text is split and billed as two, and a single non-Latin character drops the limit to 70. |
| Tweet | 280 characters | Links count as a fixed length regardless of how long the URL actually is. |
| Abstract | 150 to 250 words | Varies by journal, and is one of the few word limits that is genuinely enforced. |
| UCAS statement | 4,000 characters | Characters, not words, and spaces are included — which is why the count on this page separates the two. |
03 · reading time
That is the figure this counter uses for silent reading, and it is an average of averages. Real reading speed varies enormously with the material: a familiar news story goes faster, a legal clause or a page of statistics much slower, and anything the reader stops to think about is not really being read at speed at all.
Speaking time uses a hundred and thirty words a minute, which is closer to a prepared presentation than to conversation. People talk faster than that in a meeting and slower than that when reading a script aloud, so treat it as a planning figure rather than a measurement — useful for working out whether a talk fits its slot, not for timing it exactly.
The one thing both numbers are genuinely good for is comparison. Whether a draft is a four-minute read or a fourteen-minute read is a real difference, and it is much easier to feel from the estimate than from a word count.
04 · writing to a limit
Anyone who has written to a hard limit knows the two directions are not symmetrical. Cutting a long draft down improves it almost every time, because the first things to go are the ones that were not carrying weight. Padding a short one to reach a minimum almost never does.
So when a limit is fixed, the useful habit is to overshoot deliberately and then cut. A draft at a hundred and twenty percent of the target cuts down to something tighter than a draft written to land exactly on it, and the counter is what tells you where you are without interrupting the writing to guess.
The other habit worth having is leaving a margin. Counters disagree by a word or two over hyphenation and stray numerals, and a form that rejects a submission at four thousand and one characters does not explain which counter it used. Landing at three thousand nine hundred and fifty is not caution, it is just avoiding a problem that has no diagnostic.
Where there is no limit at all, length is a poor target in either direction. The question worth asking is whether the piece covers what it promised, and the reading time above is a better proxy for that than the raw count — a fourteen-minute read had better be answering something substantial.
05 · the frequency list
The list under the statistics strips out the common function words and shows what is left. It is the fastest way to notice that the same adjective appears nine times in a piece you thought was varied — something almost impossible to see while writing and obvious the moment it is counted.
It also answers a different question for anyone writing for search: whether a page actually covers the subject it claims to. Not as a target to hit — stuffing a word in repeatedly reads exactly as badly as it sounds — but as a check that the topic appears at all.
The density figure beside each word is its share of the total. Anything much above three percent is usually a sign of repetition rather than emphasis.
06 · questions
Q01
A word is a run of letters or digits, optionally joined by an apostrophe or a hyphen. So "don't" is one word and "twenty-one" is one word, which matches how most editors and word processors count. Numbers count as words too.
Q02
Because different limits mean different things. Twitter counts characters including spaces; some form fields do not; and a meta description limit is really about pixel width rather than either. Having both means you can check against whichever one applies.
Q03
At two hundred words a minute, which is a common average for silent reading of ordinary prose. Speaking time uses a hundred and thirty, closer to a measured presentation pace. Both are estimates: dense technical text reads slower and dialogue faster.
Q04
No. The counting happens in your browser and nothing is sent to a server. That matters more than it sounds for this particular tool, since the text people paste into a word counter is usually a draft they have not published yet.
Q05
A run of text ending in a full stop, question mark or exclamation mark. It is a good approximation and not a perfect one — abbreviations like "e.g." and decimal points inside numbers can inflate the count slightly on technical text.
Q06
It shows which words you lean on. Common function words are filtered out, so what is left is usually the vocabulary of the piece. It is a quick way to notice that you have used the same adjective nine times, which is hard to see while writing.
Q07
Yes, for anything written with letters — the tokeniser is Unicode-aware rather than English-only. The frequency list filters English function words on this page; the Spanish and Portuguese pages filter their own.
Q08
Around five hundred for a typical double-spaced manuscript page, and closer to six hundred single-spaced, though it depends entirely on the font, the size and the margins. Publishers who ask for a page count are usually asking for a word count and would rather have that.
Q09
The formatting is dropped and the text is counted as plain characters, which is what every limit you are checking against will do too. Pasting from a word processor sometimes brings in non-breaking spaces, and those count as characters just like ordinary spaces.
Q10
Usually by a word or two, over hyphenation or how a stray number is treated. No two counters agree exactly, which is worth knowing if you are writing to a hard limit — leave a little margin rather than landing on the number exactly.
No two counters agree exactly, so when a limit is enforced, land a little under it rather than exactly on it. Everything here runs in your browser and nothing is stored.