Advanced Word Counter

Word counter that counts words, characters, sentences and reading time as you type

Paste or type text into the box above and the counts update while you work. There is no submit button and no upload step — the analysis runs inside your browser, so the text never leaves the machine you are typing on. Your draft is kept in your browser’s own local storage so it survives a refresh, and the Clear button wipes both the box and that saved copy.

Most people arrive here with a specific number in mind: a 500-word assignment, a 280-character post, a meta description that has to fit, or an estimate of how long something takes to read aloud. The sections below cover what each figure actually measures, where the edge cases are, and why two word counters can disagree about the same paragraph.

What this tool measures

The Statistics tab shows every basic figure at once rather than making you switch modes:

Beyond the counts there are five more tabs: Keywords (the fifteen most frequent non-stop-words with their density, clickable to highlight), Readability (six published formulas), SEO (a length and density checklist), Social (character limits for the major platforms), and Goals (a word or time target with a live words-per-minute figure based on how long you have actively been typing).

What it does not do

There is no file upload, so counting a .docx or a PDF means opening it and pasting the text. There is no grammar or spelling check, no plagiarism check, and no export to a file — the copy button puts your text back on the clipboard, not a report. Word counts are language-agnostic in the sense that whitespace splitting works for any space-separated script, but the readability formulas and the syllable estimator are built for English and will produce meaningless numbers for Chinese, Japanese or Thai text, which does not use spaces between words at all.

Where “one word” gets ambiguous

Word counting sounds trivial until you look at real text. The rule here is whitespace-delimited: anything separated by a space, tab or newline counts as one word. That gives predictable, explainable results, and it is the same rule most writing tools use — but it has consequences worth knowing before you argue with a marker or a client about a number.

• Hyphenated words. state-of-the-art is one word here, because there is no space in it. Some style guides and some counters treat each hyphenated element separately, which would make it four.
• Em dashes. writing—quickly with no surrounding spaces is one word; writing — quickly is three, because the dash itself becomes a token. Spacing around dashes is the single most common reason two people counting the same sentence get different answers.
• Contractions. don’t and it’s are one word each. Nobody seriously disputes this, but note that the average-word-length figure strips the apostrophe, so don’t contributes a length of 4, not 5.
• Numbers and dates. 1,250 is one word. 12 March 2026 is three. A table of figures pasted in will inflate the word count far above what a reader would call “words”.
• URLs and email addresses. https://example.com/some/long/path counts as a single word but as a very long one, which drags the average word length up and can wreck the readability scores for an otherwise plain piece of text.
• Ellipses and stray punctuation. A standalone ... surrounded by spaces counts as a word. So does a lone bullet character pasted from a slide deck.
• Headings, captions and footnotes. The tool counts whatever is in the box. If your assignment excludes the bibliography and the abstract, paste only the body text.

Why counters disagree on the same text

If this tool says 1,004 and your word processor says 998, neither is broken. The differences almost always come from one of four places: how hyphens and dashes are tokenised, whether numbers and symbols count as words, whether text in headers, footers, footnotes, tables and text boxes is included, and whether a trailing element like a page number is being swept in. Word processors count what is in the document; this tool counts what is in the box. For any submission with a hard limit, the count that matters is the one produced by whatever the person receiving it will use — if the brief says “Microsoft Word count”, treat that as the authority and use this tool for the fast iterative drafting in between.

Sentence counting is looser still. Splitting on ., ! and ? means Dr. Smith, e.g., 3.5 and example.com each register as sentence breaks. In ordinary prose the error is small; in text dense with abbreviations, decimals or domains, the sentence count reads high and the average sentence length reads low. That in turn skews every readability score, since all six of them divide by sentence count.

Counting for a limit

Essays and assignments

Academic limits are usually expressed in words, with a stated tolerance. Paste the body and watch the Words figure; the Pages figure divides by 250 words per page, which is a rough double-spaced convention and not a substitute for checking the actual page count in your document at your required font and margins. If you are trimming to fit, the Keywords tab is a fast way to spot the phrase you have used eleven times, and the average sentence length is a useful proxy for where the padding sits — sentences averaging well over about 25 words usually contain cuttable clauses.

Meta titles and meta descriptions

Search snippets are truncated by pixel width, not by character count, so any character target is an approximation. The Social tab includes the two conventional working limits used here: 60 characters for a meta title and 160 for a meta description, with a comfortable band a little below each. Two practical notes. First, watch the characters figure, not words — a title of eight short words and a title of eight long ones truncate very differently. Second, capital letters and wide characters such as W and M take more pixels than i or l, so an all-caps title at the character limit will still be cut. Aim under the limit rather than at it, and put the load-bearing words first so they survive truncation.

Social posts

The Social tab holds the platform ceilings and shows a progress bar against the one you select:

The visible-before-truncation figure is usually the one that matters more than the ceiling. A LinkedIn post can run to 3,000 characters, but the reader decides whether to expand it based on the first line or two, so write to the visible window and let the rest be depth. Be aware that platforms count characters their own way — emoji, non-Latin scripts and shortened links are frequent sources of a discrepancy between a local count and what the compose box reports, so treat these numbers as a guide and confirm in the platform itself before publishing at the very edge of a limit.

Reading time

The reading-time estimate divides the word count by 238 words per minute; speaking time uses 140. Both are averages for adult English readers and speakers on straightforward material. Silent reading of technical prose, code samples or anything the reader stops to think about runs slower; skimming runs much faster. For a spoken script, 140 words per minute is a deliberately unhurried pace — useful for a presentation with a fixed slot, because rehearsing at that speed leaves room for pauses. If you are writing a five-minute talk, roughly 700 words is a sane starting target and then you cut.

Readability, keywords and the SEO checklist

The Readability tab computes six standard formulas: Flesch Reading Ease, Flesch–Kincaid Grade Level, Gunning Fog, SMOG, Coleman–Liau and the Automated Readability Index. All of them are functions of sentence length combined with either syllables per word or characters per word, so they measure surface complexity, not whether an argument is clear. Two caveats specific to this implementation: syllables are estimated by counting vowel groups with an adjustment for a silent trailing e, which is a good approximation but wrong on words like fire or business; and SMOG is only meaningful with at least 30 sentences, so with shorter text the tool falls back to the Gunning Fog value rather than printing a number the formula cannot support.

The Keywords tab lowercases the text, strips punctuation, drops a built-in stop-word list plus anything two characters or shorter, and reports the fifteen most frequent remaining terms with their density as a percentage of that filtered set. Because stop words are excluded from the denominator, the density figures here run slightly higher than a naive count over all words — useful for comparison between terms in the same document, less useful compared against a number from a different tool. Clicking a keyword highlights every occurrence in your text, which is the fastest way to see whether a term is spread through the piece or clustered in one paragraph.

The SEO tab is a checklist rather than an audit: it flags whether the text passes 300 words, whether it sits in a longer band, whether the Flesch score falls in a comfortable range, and whether the top keyword’s density is in the low single digits. Treat those thresholds as conventions the industry uses, not as ranking rules — they are handy guard rails against publishing something visibly thin or visibly stuffed.

Privacy and how it runs

Everything on this page is computed in JavaScript in your browser. There is no request to a server carrying your text, no queue, and no copy retained anywhere else — which matters if what you are counting is an unpublished manuscript, a client document or anything under NDA. The one thing that is stored is a copy of your draft in your own browser’s local storage, so closing the tab and coming back does not lose your work. That copy is per-browser and per-device; it is not synced, and clearing site data or pressing Clear removes it. On a shared or public computer, press Clear when you finish.

Because the work is local, very long documents are limited by your device rather than by an upload cap. Analysis is debounced and the heavier readability and keyword passes are deferred, so typing stays responsive; on a book-length paste, expect a brief moment before the derived tabs settle while the basic counts appear straight away.