Count the exact text a form will receive, using the unit its instructions name. A word limit and a character limit answer different questions, and the destination may count either one differently from another editor.
Word Counter reports words, characters, characters without whitespace, code units, code points, sentences and paragraphs. Its main character count uses grapheme clusters in browsers with the required segmentation support. This is closer to counting visible characters than measuring the length of a JavaScript string.
Find the receiving form's rule
Read the instructions beside the field. Does the limit include spaces, headings, references or line breaks? For separate fields, count each answer separately. Include a heading only if it will be pasted into that same field.
If the form has its own counter, test the final text there. Its acceptance rule controls the submission, even when another counter gives a different total. A small margin can help while you edit, but it does not replace checking the destination.
Understand differences around emoji and words
A single visible emoji can contain several code points and still more UTF-16 code units. An accented letter can also be stored as one character or as a letter plus a combining mark. Word Counter shows the measurement method and additional totals so you can see why two counters disagree. In browsers without Intl.Segmenter, its main count falls back to code points.
Word matching keeps internal apostrophes and hyphens in ordinary letter or number sequences. Where supported, it also segments matched Han, Japanese and Korean text into words. Those rules are estimates for a particular implementation, rather than a universal linguistic definition. Check the destination especially carefully for languages and punctuation conventions that it may handle differently.
The MDN explanation of Intl.Segmenter describes the distinction between grapheme, word and sentence segmentation.
Trim a draft without losing its point
For a hypothetical 250-word statement measured at 288 words, you need to remove at least 38 words under that same counting rule. Start with a repeated explanation or an example the question does not need. Recount after editing, then read the answer as a whole.
Use Text Compare to compare the full and shortened versions when you need to confirm that a qualification or fact survived. A frequent-word list can reveal repetition, but deleting a common word everywhere is not a useful editing rule.
Copied text can contain tabs, unusual spaces and accidental line breaks. Remove Extra Spaces and Line Breaks can change those characters. Count the cleaned version only if that is the version you intend to submit, and review any joined lines.
Treat reading time as an estimate
The reading-time setting uses 150, 200 or 250 words per minute. For example, 300 words at 200 words per minute gives about 90 seconds. Technical terms, pauses and the reader's familiarity can change the real time. Time a script aloud when duration matters.
The input limit is 2,000,000 UTF-16 code units. The page's short-writing guides are drafting aids, including its search-title and description targets. They are not promises about platform acceptance or the exact snippet a search engine will display.