How it works
- You paste or type into the box. The text stays in that field — it is never sent to a server, written to storage, or included in an analytics event.
- Words are counted by trimming the text and splitting on runs of whitespace, which is the definition that matches what a word processor reports for ordinary prose. Characters are counted as JavaScript string length, so an emoji built from a surrogate pair counts as two.
- Sentences are counted by matching runs of terminal punctuation — full stops, question marks and exclamation marks. Consecutive marks count once, so an ellipsis does not inflate the total.
- Reading time assumes 220 words a minute, rounded up to at least one minute. That figure is a common estimate for adult silent reading of general prose; technical material is slower and it is worth treating the number as a floor.
- The whole routine runs on each keystroke in plain JavaScript, so the counts update as you type. Close the tab and the text goes with it.
Limitations
It counts on every keystroke, and above roughly two million characters that stops keeping up on a mid-range phone. The counting itself is fast; re-running it on every input event for a document that size is what falls behind. Split a very long text in half and count each part.
Whitespace splitting is the wrong definition for some languages. Chinese, Japanese and Thai do not put spaces between words, so a paragraph in any of them reports as one or two words regardless of length. Character count remains meaningful there and is the figure to use.
Hyphenated compounds count as one word and an em dash between two words counts as none, which is what a word processor does but not always what an editor means. Abbreviations with internal full stops inflate the sentence count, so a paragraph dense with 'e.g.' and 'i.e.' will read as having more sentences than it does.
Reading time is an average applied to your specific text. A dense contract clause and a chatty blog paragraph of identical length do not take the same time to read, and neither takes the time this reports if the reader is skimming.
Questions
Is my text uploaded?
No. The text stays in the field and is processed in this tab. Nothing is posted, logged or kept after you close it — which is the point, since the things people count most often are drafts that are not public yet.
What are the limits?
Counting keeps up comfortably into the hundreds of thousands of characters. Above roughly two million it starts lagging behind your typing, because the count re-runs on every keystroke. Split the text and count each half.
Does the count match Microsoft Word?
For ordinary prose, yes — both split on whitespace. Small differences appear around hyphenated compounds, footnote markers and text in tables, and Word counts headers and footers depending on the setting.
How is reading time calculated?
Words divided by 220, rounded up to at least a minute. It is an average for adult silent reading of general prose. Technical or unfamiliar material runs considerably slower, so use it as a lower bound rather than a promise.
Do I need an account?
No. There is no sign-up, no email step and no usage counter. The tool is paid for by the ads on this page.
When people use it
Most people arrive with a limit to hit and text already on the clipboard: an application essay capped at five hundred words, a meta description that has to fit, a bio for a conference programme, an assignment with a stated range. The count updates as you cut, so trimming happens against live feedback instead of a guess and a re-check.
The second use is estimating. A writer quoting for a piece, or someone deciding whether an article is worth the next ten minutes, wants a rough duration rather than an exact count. Pasting the draft in and reading the reading-time figure answers that faster than opening a word processor, and it does not require the document to exist as a file.
An essay written to a word limit is usually submitted through a portal that also caps the file size, and a PDF full of scanned figures will miss it. Compress PDF re-encodes the images inside the document without touching the text you just counted.
How it compares
A word processor counts too, and counts more precisely across tables, footnotes and headers. What this does that a word processor cannot is count something that is not a document yet — a paste from a chat, a form field, a half-written email — without creating a file. And unlike the online counters that post the text to a server to count it, nothing here leaves the tab, which matters for a draft under embargo.