The counter says my word count is approximate in this browser; what does that mean and can I trust it?

It means your browser does not have Intl.Segmenter, the built-in function that splits text into characters, words and sentences under the full Unicode rules, so the tool uses a simplified rule instead. Code points, bytes, lines and paragraphs stay exactly right. Characters, words and sentences become approximate and can slip with combined emoji and with languages written without spaces. For ordinary Arabic or English text they usually match, but if the figure will go on an invoice or a delivery, open the page in a current browser.

Published · 3 min read

Which numbers are affected

  • Always exact: code points, UTF-16 units, UTF-8 bytes, lines, non-empty lines and paragraphs. These are worked out in a way that never needs the missing function.
  • Approximate: characters as read, characters without spaces, words and sentences.

The message appears under the counts in the Word and character counter, and is also added to the result you copy or download, so an approximate figure never reaches someone else without a warning.

Where the approximation really slips

We ran the tool's logic on various texts twice, once the full way and once with the same simplified method an older browser gets, and compared:

  • An Arabic welcome line with a number and an exclamation mark: 11 words, 57 characters and 2 sentences both ways. A perfect match.
  • A text with the joined family emoji and fire emoji: identical figures both ways.
  • Chinese with no spaces, «我们的商店欢迎您»: 5 words the full way, 1 the simplified way, because it relies on spaces and punctuation.
  • English e.g. apples: one sentence the full way, two the simplified way.
  • «نعم… وصلت الطلبية» (yes… the order arrived): one sentence the full way, two the simplified way, because of the ellipsis.

What these tests show: in ordinary Arabic or English text, words and characters matched, and the differences were in sentences around full stops and ellipses, and in words for languages written without spaces. These are tests on specific examples, not a guarantee for every text.

What to do in practice

  1. If you need a rough figure for yourself, such as whether an article is near 800 words, the current result is usually enough.
  2. If the figure goes into a quote, an invoice or a client delivery, open the page in a current version of a mainstream browser and count again.
  3. If you are on an old device whose browser cannot be updated, rely on bytes and code points when the limit you face is technical, since those are exact.
  4. If your browser shows this message even though it is up to date, check you are not in a browser built into another app, as some of those are older than they look.

Why the tool does not hide this

The tool could show the approximate figure without comment, and most people would never notice. But a counter used to price translation and copywriting should say when its figure is less exact. The message does not mean the result is wrong, only that it was not worked out with the full rules.

Common questions

Does the simplified method change anything for privacy?

No. Both methods run inside your browser, and the text is sent nowhere either way.

Does the message appear in current browsers?

It should not, since they include the function. If it appears, that browser does not provide it, whatever the reason.