Word and character counter.

Words are counted the way Microsoft Word and Google Docs count them: runs of text separated by spaces or line breaks that contain at least one letter or digit. A hyphenated word such as well-known or e-mail counts as one, and a dash standing on its own counts as none. Chinese, Japanese and Thai are written without spaces, so for those the browser's own word segmenter (Intl.Segmenter) splits the text into words.

Characters are counted as you see them (grapheme clusters), so an emoji with a skin tone or an accented letter typed as two code points counts as one. Without spaces leaves out spaces, tabs and line breaks. Numbers such as 2026 or 3.5 count as words, as they do in Word. Sentences come from the browser's sentence segmenter, and paragraphs are non-empty lines.

Reading time uses 238 words per minute, the average silent reading speed for adult non-fiction in Marc Brysbaert's 2019 review of 190 studies. Speaking time uses 150 words per minute, a typical pace for presentations and podcasts. Both are estimates: technical text reads slower, and a nervous speaker talks faster. For video and podcast scripts, speaking time is the better guide; leave room for pauses, slides and questions.

Blogs often show reading time near the title. Rounded up to the next minute, it sets expectations and tends to reduce bounces from readers who wanted something shorter.

X counts characters with weights: most Latin, Greek, Cyrillic, Arabic and Indic characters count 1, while Chinese, Japanese and Korean characters and emoji count 2, and every link counts 23 regardless of length. A standard post allows 280 weighted characters; Premium subscribers can post up to 25,000, though the timeline collapses long posts after the first 280.

LinkedIn posts allow 3,000 characters, with only about the first 210 shown before "see more", so lead with the point. Threads posts allow 500 characters (longer text needs an attachment), Bluesky 300 graphemes, and Instagram captions 2,200, with roughly 125 visible before "more". Platforms change these limits, so treat them as current guidance rather than guarantees. Line breaks and the spaces around hashtags and mentions count too, so check a post that is close to the limit in the platform's own composer before scheduling it.

There is no character limit for meta descriptions. Google truncates snippets by pixel width, which works out at roughly 150 to 160 characters on desktop and less on mobile, and it often rewrites the description to match the query anyway. Aim for about 150 characters with the key point first, written for people: a clear summary of what the page offers earns more clicks than a list of keywords. Title tags are cut at around 580 to 600 pixels, roughly 60 characters.

These two bars turn amber rather than red when you go over, because they are soft limits. For the wider picture, see technical SEO or our SEO services.

A single SMS carries 160 characters in the GSM-7 alphabet. Longer messages are split into parts of 153 characters, because each part needs a header to be reassembled. Characters from GSM-7's extension table (^ { } [ ] ~ \ | €) take two slots.

One character outside GSM-7, such as an emoji, a curly quote pasted from a document or most non-Latin letters, switches the whole message to UCS-2, where the limits drop to 70 characters, or 67 per part. Providers bill per segment, so a single ✅ in a one-time-password template can double its cost. Some providers swap common Unicode characters such as curly quotes for GSM-7 equivalents before sending, but check your provider's settings rather than relying on it. The SMS panel shows the encoding, the segment count and the exact characters that forced UCS-2. For rich, longer messages, the WhatsApp Business API is the usual alternative.

Many writing tasks come with a budget. Essays and grant applications set word limits and may or may not count headings, references and footnotes, so check the rules and paste only the parts that count. App store descriptions, push notifications, ad headlines and form fields set character limits, and a translation can run much longer than the English source, so leave headroom.

Counts also vary slightly between tools: some split on hyphens or slashes, some treat URLs and email addresses differently. This counter follows the Word and Google Docs convention, so its numbers line up with what an editor or a client is likely to check.

For text sent to an AI model, words are the wrong unit: models count tokens, and the ratio of tokens to words changes with language and content. Use the LLM token estimator for prompts and context windows.

Keyword density is how often a word or phrase appears as a share of all words. Search engines have not rewarded a target percentage for many years, and stuffing a phrase reads badly. Use the list to spot accidental repetition and to confirm the topic you meant to cover is actually the one you wrote about. Two-word phrases are usually more revealing than single words. Readability matters more than density: short sentences, concrete words and one idea per paragraph keep people reading.

Common words such as the, and and of are skipped by default; untick Skip stop words to include them. Everything is computed in your browser, so drafts, contracts and unpublished posts stay private. Our content marketing team uses the same checks before publishing.