Every word sent to GLOSSA counts, whatever it was sent for. Reused translations and anything your team writes by hand do not.
Who this is for
Anyone watching the meter.
What happens next
Counts: the source words of anything handed to GLOSSA. Translating a string is the obvious one. These count the same way, against the same allowance, the same hard limit and the same invoice:
Quality checks. Scoring a translation hands it to GLOSSA to read, so the source words of every string in the check count.
Brand voice drafts. Asking GLOSSA to draft a style guide from translations you have already approved counts the words it was given to read.
Repairs. A string that fails a check and is translated again counts its source words again, on every try. Auto-heal does this without anyone pressing a button, and it counts the same in suggest mode, where the new wording is attached for a reviewer instead of being written in.
Does not count:
Reused translations. When the exact same source text has been translated before, transglot reuses that translation at no cost instead of translating it again. That is what the Words reused at $0 figure on your dashboard is measuring, and it is why turning translation memory off makes your usage go up.
Anything you or your team writes by hand.
Reading, filtering, approving.
The scorecard, which is a transform over findings the quality checks have already produced and hands GLOSSA nothing of its own.
Three practical consequences:
Turning auto-translate on means new and changed source strings are translated as they arrive, without anyone pressing a button. That is usually what you want, and it is also the setting most likely to surprise you on the meter.
Adding a language multiplies the work: every untranslated key in it is a translation to pay for.
Keeping translation memory on is the single biggest lever on the bill.
The words are the account's, shared by every workspace in it. The account's Usage page, which its Owners and Billing members open, shows the words used this cycle against what the plan includes, and By workspace shows which workspaces used them.
Related
Usage-based billing
Translation memory
AI translation settings