Word & Character Counter
Real-time, local text metrics. Your inputs are never transmitted over the network.
Document Statistics
Time Estimates
Readability
Flesch Reading Ease
46.0
Difficult (college)
Flesch-Kincaid Grade
9.8
High school
Gunning Fog
12.5
College
SMOG Index
11.8
Needs 30+ sentences to be meaningful
Coleman-Liau
13.7
College
Automated Readability
10.1
High school
146 syllables · 11.9 words per sentence · 1.76 syllables per word · consensus grade 11.6
Social Limits
Top Keyword Density
Top Character Density
How is Reading and Speaking Time Computed?
Our reading calculator assumes an average reading speed of 200 words per minute (WPM), which is standard for adults reading informational copy. For speaking, we assume an conversational standard speed of 140 words per minute.
Six Readability Formulas, Not One
The readability panel reports Flesch Reading Ease, Flesch-Kincaid Grade, Gunning Fog, the SMOG Index, Coleman-Liau and the Automated Readability Index, plus a consensus grade averaging the five grade-level scores. They disagree by design — each weighs sentence length, syllables and word length differently — so a consensus is more informative than any single number. SMOG is labelled unreliable until your text reaches 30 sentences, because that is the sample size it was validated on. Syllable counting is a heuristic, not a dictionary.
Stopword Filtering for Accurate Content Density
Unlike basic calculators that simply count every word, ZeroServer's Word Counter excludes common English stopwords (such as "the", "and", "or", "to"). This ensures that your keyword density analysis reveals the actual semantic content and core keywords of your text, which is invaluable for SEO optimizing, drafting metadata, or editing blogs. Counting is Unicode-aware, so Cyrillic, Greek, accented Latin and CJK text are tallied rather than silently discarded.
Built and maintained by Meet Shah · Last updated
What this tool is used for
- Checking a draft against a word limit before submitting it.
- Measuring a section's length to decide whether it needs splitting.
- Counting characters for a field with a hard cap.
- Checking which terms a piece emphasises.
- Comparing two drafts on length rather than impression.
How it works in practice
A worked example
A caption has to fit a platform limit, and the character count here disagrees with the one the platform itself shows.
naïve café 👨👩👧👦 family 日本語のテキストです
Characters (spaces) Words naïve café 10 2 👨👩👧👦 family 8 1 日本語のテキストです 10 4 ← segmented
The middle row is where the disagreement lives. That family emoji is a single thing a reader sees and eleven UTF-16 units of storage, joined by invisible zero-width characters, so the count here is 8 while a naive length check in most languages reports 18. This page counts what a person perceives, which is the right answer for a caption and the wrong answer for a database column. The bottom row is Japanese, which has no spaces to split on — the segmenter finds four words where a reader would say five.
The edge case that catches people
So there is no single number called the character count. This page counts perceived characters; a database column counts bytes or code units depending on the engine; a text message counts seven-bit units and drops to sixteen-bit ones the instant a single character falls outside the basic set, roughly halving the message; and at least one social platform weights ranges differently so that Latin text costs half of what everything else does. Four honest answers for one string, and the only way to know which applies is to know what is measuring at the far end.
When not to use this tool
As a limit check it is useful and as a quality signal it is nothing. Length is the one property of writing that is trivial to measure and almost never the property that matters, so a target hit by padding is worse than a draft that fell short. The reading-time figure carries the same caveat twice over: it assumes silent reading of ordinary prose at an average adult pace, which is the wrong model for anything read aloud, anything technical and anything with code in it — a script wants roughly two-thirds of that speed.
Frequently Asked Questions
- How are words counted?
- Words are split on whitespace (spaces, tabs, newlines). Hyphenated compounds count as one word. Attached punctuation is stripped before counting. The algorithm matches the convention used by most word processors.
- How is reading time calculated?
- Based on 238 words per minute — the average adult silent reading speed from peer-reviewed research (range: 200–300 wpm). The estimate assumes average familiarity with the content type.
- What are the social media character limits shown?
- Twitter/X: 280 characters. LinkedIn post: 3,000 characters. Instagram caption: 2,200 characters. SMS: 160 characters (single segment). These limits are shown as reference bars so you know when you're approaching them.
- Why might my count differ from Word or Google Docs?
- Different apps count contractions, hyphenated words, and code differently. Some count 'don't' as one word, others as two. This tool uses whitespace splitting — the most common convention for plain text.
- Can I count words in non-English text?
- Yes for most scripts that use spaces between words. CJK (Chinese, Japanese, Korean) text may undercount — those languages don't separate words with spaces. For CJK, character count is typically more meaningful than word count.
Common errors and gotchas
- Assuming every tool counts the same way, since hyphenated words and numbers are treated differently.
- Counting characters without deciding whether spaces are included, which most limits do count.
- Counting code blocks and captions as prose against a prose limit.
- Using a character count as a proxy for a byte limit, which differs for non-ASCII text.
- Treating a word count as a quality measure rather than a constraint.