Word Counter

How to use the word counter

Type or paste your content into the box. Every metric updates instantly with no button to press. The platform-limit bars fill as your character count approaches each platform's cap — green below 75%, amber at 75–90%, red above 90%.

What is counted

• Words — sequences of non-whitespace characters separated by spaces, tabs or newlines. Numbers and hyphenated words count as one word each.
• Characters (with spaces) — every Unicode codepoint including spaces and punctuation. This is the number most social platforms use for their character limits.
• Characters (no spaces) — codepoints excluding all whitespace. Useful for SEO character budgets and academic character targets.
• Sentences — runs of text ending in ". ! ?" followed by a space or end-of-input.
• Paragraphs — blocks of text separated by one or more blank lines.
• Reading time — word count ÷ 200 words per minute (average adult web-reading pace).

Platform character limits

What actually counts as a "word"?

Almost every word counter on the web — this one included — defines a word as a run of non-whitespace characters bounded by spaces, tabs or line breaks. This is called whitespace tokenisation, and for English and other space-separated languages it matches what people intuitively mean by "word". It does have edge cases: a hyphenated compound such as well-being, a contraction such as don't, a URL, an email address and a number such as 3.14 are each counted as a single token, because none of them contains an internal space.

Whitespace tokenisation falls apart entirely for languages that do not put spaces between words. Chinese, Japanese and Thai — along with Khmer, Lao and Burmese — are written as continuous strings of characters with the word boundaries left for the reader to infer. A forty-character Chinese sentence might contain a single space or none at all, so a whitespace-based word count would report it as one or two "words", which is meaningless. Splitting those scripts into real words takes a dictionary or a statistical segmenter, not a simple space split. For that reason the meaningful length metric in CJK and Thai text is the character count, and platforms adjust accordingly: Medium, for example, switches from words-per-minute to roughly 500 characters per minute when it estimates reading time for Chinese, Japanese and Korean. If you are working in one of these languages, trust the character figure on this page rather than the word figure.

Reading time vs speaking time: what the research says

This tool estimates reading time at 200 words per minute, a deliberately conservative pace that suits skim-friendly web copy. The most thorough modern evidence comes from Marc Brysbaert's 2019 meta-analysis "How many words do we read per minute?" in the Journal of Memory and Language, which pooled 190 studies covering 18,573 participants. It found that adults reading English silently average about 238 words per minute for non-fiction and 260 wpm for fiction, with most readers falling between 175 and 320 wpm. Reading aloud is slower — roughly 183 wpm — because speech production becomes the bottleneck. Children, older adults and people reading in a second language are slower still, so a single universal figure can only ever be an estimate.

Speaking is slower than silent reading because the mouth cannot keep up with the eye. Conversational English runs around 150 wpm; the American Speech-Language-Hearing Association's widely cited benchmark for a clear presentation is closer to 130 wpm; audiobook narrators and podcasters typically sit at 150–160 wpm; and TED speakers average about 163 wpm. If you are writing a script, divide your word count by your target speaking rate rather than a reading rate — a 1,500-word talk at 130 wpm is roughly 11–12 minutes, not the 7.5 minutes a reading estimate would suggest.

These figures are population averages, not hard limits; an individual's pace varies with the material's difficulty, familiarity and purpose. Treat any reading-time badge — here or on a publishing platform — as a friendly approximation rather than a measurement.

Character limits beyond social posts: SMS, search and ads

The limit bars above cover the major social networks, but several other contexts impose their own character budgets — and some of them behave in surprising ways.

Text messages are the classic trap. A single SMS using the standard GSM-7 alphabet holds 160 characters; the moment you include one character outside that alphabet — most emoji, many accented letters, or any non-Latin script — the whole message switches to UCS-2 (16-bit) encoding and the limit drops to 70 characters. Longer messages are split into segments and stitched back together, which spends part of each segment on a header: concatenated GSM-7 messages carry 153 characters per segment and UCS-2 messages 67. So a 75-character message full of emoji can cost you two segments where the same length of plain text would cost one.

Search snippets are measured in pixels, not characters. Google renders a page title up to about 600 pixels wide on desktop — usually 50–60 characters — and a meta description up to roughly 920 pixels, around 155–158 characters (closer to 120 on mobile). Because wide letters such as W and M eat more pixels than i or l, two titles of identical character length can truncate differently. Google Ads is stricter and counts characters directly: each responsive-search-ad headline is capped at 30 characters, each description at 90, and double-width CJK characters count as two.

Platform limits change without much warning — X raised the post limit from 140 to 280 in 2017 and now lets paid accounts post up to 25,000 characters — so always confirm against the platform's current documentation before a high-stakes publish.

Why an emoji can count as two (or more) characters

If you have ever pasted a single emoji into a form and watched the character counter jump by two, you have met one of the deepest quirks in modern text handling: a "character" is not one thing. There are three different units that can all reasonably claim the name, and they routinely disagree.

• UTF-16 code units — what JavaScript's string.length actually returns. Each unit is 16 bits and can address values up to U+FFFF.
• Unicode code points — the logical characters in the standard, numbered from U+0000 up to U+10FFFF. Anything above U+FFFF (most emoji, many rare scripts) does not fit in one 16-bit unit, so JavaScript stores it as a surrogate pair of two code units. The grinning face U+1F600 is stored as the pair 😀 and so reports a .length of 2.
• Grapheme clusters — what a human sees as one character, defined by Unicode Standard Annex #29 ("Unicode Text Segmentation"). A grapheme can be built from several code points.

The gap between these layers grows fast. A national flag such as the United Kingdom's is two regional-indicator code points (U+1F1EC plus U+1F1E7), which is four UTF-16 code units but a single grapheme. A waving hand with a skin tone is the base emoji plus a modifier code point. A family emoji is assembled by gluing several people together with the zero-width joiner U+200D — the man-woman-girl family is five code points, and adding a second child pushes it to seven, yet it is still one grapheme you would count as a single character.

This counter reports characters by code point, which is the unit most social platforms use for their limits, so its numbers line up with what X or Instagram will accept. If you need the human-visible count instead, modern browsers expose Intl.Segmenter with grapheme granularity, and iterating a string with the spread syntax [...text] walks it by code point rather than by code unit.

Word counts for writers: manuscript pages and document lengths

Publishing and academia still think in pages, and the bridge between a word count and a page count is a long-standing convention. The traditional estimate is 250 words per double-spaced page in 12-point type with one-inch margins — a figure that dates back to monospaced typewriter fonts such as Courier. Modern proportional fonts fit considerably more on the same physical page, so a literal word-processor document at those settings can run closer to 300–400 words per page; editors and agents keep using the 250-word rule anyway because it gives a stable, font-independent way to estimate length. Most US and Canadian publishers quote a guideline of 250–300 words per manuscript page.

Rough word-count norms are worth knowing when you plan a piece, though every figure varies by genre and publisher:

• A standard double-spaced manuscript page ≈ 250 words.
• A typical 500-word blog post ≈ 2 manuscript pages, about two and a half minutes' reading.
• College essays are now usually set in words (for example 1,500–2,500) rather than pages.
• Flash fiction runs under ~1,000 words; a short story is roughly 1,000–7,500.
• Most adult novels land between 70,000 and 100,000 words, with 80,000–90,000 a common target.

Because these ranges are conventions rather than rules, use the live count above to hit a specific target and treat any page estimate as approximate.

Keyword density, stuffing and readability

Writers optimising for search often ask what "keyword density" — the percentage of words that are a target phrase — they should aim for. The honest answer is that there is no target number. Google's own spam policies treat keyword stuffing (repeating words unnaturally to manipulate rankings) as a violation that can hurt rather than help a page, so the goal is natural phrasing, not a ratio. A word counter is still useful here: it lets you sanity-check that a 1,200-word article is genuinely substantial, and that a phrase you have leaned on does not appear so often that the prose reads awkwardly.

A more constructive metric is readability. The best-known measure is the Flesch Reading Ease score, devised by Rudolf Flesch in 1948, which rewards short sentences and short words. Its formula is:

206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words)

The two inputs are average sentence length (total words divided by total sentences) and average syllables per word. Scores run from 0 to 100: a result of 90–100 reads very easily, around 60 is plain English a typical 13- to 15-year-old can follow, and below 30 is dense, academic prose. The sentence and word counts on this page give you the first half of that calculation directly — a quick way to see whether your average sentence has crept past the 20-word mark where readability usually starts to slide.

Frequently asked questions

How are words counted?

Words are sequences of non-whitespace characters separated by spaces, tabs or newlines. Numbers ("123"), hyphenated words ("well-being") and contractions ("don't") each count as one word. Empty lines and leading/trailing whitespace do not add words.

How is reading time calculated?

Reading time uses 200 words per minute, which is the average adult silent-reading pace for web content. Academic research (Trauzettel-Klosinski & Dietz, 2012) puts the mean at 200–300 wpm; 200 wpm gives a comfortable estimate that does not undercount.

Does character count include spaces?

Both counts are shown. "Characters (with spaces)" is the total codepoint count, which is what most platforms (X, Instagram, LinkedIn) use for their character limits. "Characters (no spaces)" excludes whitespace, which is useful for SEO character budgets where spaces do not contribute to keyword density.

Is my text sent to a server?

No. All counting runs in your browser with JavaScript. Your text never leaves your device.

Why do sentences sometimes seem undercounted?

Sentence detection splits on ". ! ?" followed by whitespace or end-of-input. Abbreviations like "Dr." or "U.S.A." inside a sentence can look like sentence boundaries — this is a known limitation of punctuation-based counting. For a precise count in a final document, always do a manual review.

Related tools

• Case Converter
• Text Diff Checker
• JSON Formatter & Validator
• URL Encode / Decode