Why word and character counts can disagree
A word is not simply everything between two spaces. Chinese and Japanese text can contain words without spaces, while punctuation and contractions need their own rules. This counter uses the browser’s language-aware word segmentation when it is available. Numbers may count as words. An institution or publishing platform may use a different rule, so use its counter for a final submission check.
The character total counts grapheme clusters: the groups of Unicode characters that users generally perceive as one character. An emoji or a letter with a combining accent may contain several code points. UTF-16 code units and UTF-8 bytes answer different technical questions. A form that limits bytes can reject text that fits a visible-character limit.
Reading time is a planning estimate, calculated as words divided by words per minute. The default pace is editable. Technical passages, pauses and unfamiliar vocabulary all change the time you actually need. Read an important speech aloud once instead of treating a calculated duration as a promise.
Repeated words can help you notice an awkward sentence, but there is no ideal keyword percentage to chase. Keep names and technical terms where they help the reader. If your text came from a scanned PDF, check OCR mistakes before counting: a broken word or a repeated page header can distort the result.