A returned file can report as complete and still carry English, because a segment that was skipped often gets pre-filled with its own source text. This separates genuinely translated units from empty ones and from those identical to the source, and lists the last group so you can judge each one.
Free · No account · Your files never upload
Version 1.2 or 2.0. Read on your computer and never uploaded, which matters when the file is a client’s unreleased course.
Sorts every unit in a returned XLIFF into genuinely translated, still empty, or identical to its source, and lists the last two so you can act on them.
Identical. A target that exactly matches its source is sometimes correct: a product name, a number, an acronym, a word the two languages share. Often it means the segment was skipped and the tool pre-filled the target with the source, which most software then counts as translated. That is how a file arrives reporting a hundred percent complete and ships English into production.
Separating that category out is the whole point. It cannot be decided automatically, because both cases look identical in the file, so every one is listed for a human to glance at. On a real course that is usually a two-minute job that prevents an embarrassing one.
It tells you the file is complete, not that the translation is good. That judgment needs somebody who reads the language, and no tool substitutes for it.