Handling source updates in translation projects

Source updates can get messy very fast. One minute everything is on track, and the next you’ve got new content, deleted segments, words get changed or added, and suddenly your translation project feels chaotic and linguists are confused.

This guide shows how to keep things smooth, protect your translators’ work, and make sure payments stay accurate.

Why source updates are tricky?

Source updates aren’t just annoying. They can really mess up your project if you aren’t careful.

Common problems include:

  • Losing translator progress

  • Creating TM inconsistencies (wrong suggestions or break)

  • Messing up word counts and payments

  • Losing existing translations

Most of this can be prevented with the right approach.

Best way to handle updates

Use this approach whenever you can:

  1. Let linguists finish the current file.

    Keep the chain of responsibility in mind. Try not to interrupt linguists mid-task and avoid rushing changes unless it is necessary.

  2. Import the updated content as a new file or version.

  3. Run pre-translation from TM.

  4. Assign linguists to update only their own segments.

  5. Recalculate word counts based on the new version.

Why this works?

  • Keeps everyone’s work safe.

  • Makes changes easier to track.

  • Makes word counts and payments simple.

  • Reduces stress for you and your team.

But what to do when updates can’t wait?

Sometimes you can’t avoid updating an active project. In case of mid-project changes:

Do this first:

  1. Ask linguists to pause their work.

    You can send a broadcast message or remove their assignments to make sure they stop working.

  2. Make sure PMs backed up XLIFF files (just to be on the safe side).

    To make your work easier you can use a template to automate XLIFF export.

    You can also run the template’s automated actions before reimporting. This helps confirm all segments and saves them in a translation memory (TM) as a backup—just in case a linguist hasn’t confirmed them yet.

Then:

  1. Import the updates as a completely new file.

    If importing as a new file isn't possible, try reimporting the whole thing.

  2. Run X-translate and pre-translation.

  3. Double-check TM matches and context.

Before restarting work:

  1. Confirm assignments.

  2. Make sure everyone knows what changed.

What to avoid?

These are the biggest sources of problems:

  • Reimporting files while linguists are still working.

  • Ignoring comments, notes, or segment status.

  • Mixing multiple rounds of updates in one file.

Make sure to follow all the above because uncoordinated reimports = lost data + chaos + extra cleanup.

What can help keep the process?

Features that make updates easier:

If possible:

  • Use a TMS/CMS connected to a repository.

  • Reduce manual file handling.

Keeping word counts and payments correct

Updates make tracking work harder. Here’s what helps:

  • Always regenerate stats after updates,

  • Use pre-translation and weighted analysis,

  • Keep both old and new versions for reference,

  • Compare files if needed (Excel works well).

Try keeping versions separate. It makes billing much easier.