Paste or write
The editor accepts articles, essays, emails, LinkedIn drafts, meta descriptions and product copy. Statistics update locally as you type.
Method & limits · reviewed 29 July 2026
What this word counter counts—and what it only estimates.
The dashboard runs in your browser, but different editors can still disagree about words, sentences and reading time. These are the exact rules used here.
The visible workflow
Choose the mode
Blog, SEO, LinkedIn, email, essay and meta modes set editable working ranges. They are drafting aids, not platform rules or evidence that a page is complete.
Review the dashboard
Track words, characters, sentences, paragraphs, reading time, speaking time, repeated terms and a transparent heuristic draft check from the same screen.
Copy the result
Use Copy stats to move a compact summary into your brief, content checklist, classroom note or editorial QA document.
Word and character rules
Letters, numbers and joins
A word contains Unicode letters or numbers. Internal apostrophes and any number of internal hyphens stay inside one token: don’t, l’esprit and mother-in-law each count once. Leading or trailing punctuation does not.
URLs and emoji
A token beginning with http://, https:// or www. runs to the next space and counts once. Emoji and standalone symbols count as characters but not words.
Spaces and line breaks
Repeated spaces and blank lines do not create words. Character count includes every character; “no spaces” removes all whitespace before counting.
Sentence and paragraph estimates
A sentence is a non-empty text run ending in a full stop, question mark or exclamation mark, plus a final unfinished run. Abbreviations, decimal numbers, ellipses, dialogue and code can split differently from a human edit. Every non-empty line counts as a paragraph.
Reading and speaking time
Reading time is words ÷ 225; speaking time is words ÷ 150. Both round up to a whole minute and show a minimum of one minute for non-empty text. They are planning estimates, not accessibility or performance measurements.
Repeated terms and language limits
Repeated-term analysis lowercases tokens, removes English and French stop words, ignores tokens shorter than three characters and shows up to eight terms appearing at least twice. Automatic language choice is a small English/French heuristic. The counter can recognise Unicode letters, but it does not segment continuous Chinese, Japanese or Thai text reliably.
The draft check is not a quality score
The 0–100 check starts from the selected working range and subtracts for long average sentences, dense paragraphs and the most repeated term. It does not evaluate truth, originality, grammar, citations, search quality or assignment compliance. Change the target and the check can change without the prose becoming better.
Privacy boundary
Analysis runs in JavaScript in the current tab. The tool code has no account, draft upload, backend storage, analytics call or advertising script. The page separately requests web-font files from Google Fonts. Copy stats writes only when you press the button; Clear empties the editor.
Verification record
On 29 July 2026, automated Chromium checks passed for live target updates, exact and over-target states, French stop-word filtering, all six writing modes, character-based meta mode, apostrophes, multi-hyphen terms, URLs, numbers, emoji, time formulas and long-input handling. Empty input and narrow mobile layout were also checked without console errors. These tests verify the published rules; they do not force Microsoft Word, Google Docs or a school rubric to use the same rules.
Report a difference
The tool is maintained by Steve T.. Send a minimal non-sensitive sample through the Pink Banana Tools contact page and include the result you expected. Do not send private client or student documents.