Skip to content
Insights

One year in production, two invoices: what 27 markets taught us

What a year of production telemetry from a global FMCG brand says about the real cost of translation, and where the money actually goes.

Lingrail Team · Shiftum1 min read

This is an example post. Replace it with a real one, or delete the file.

For twelve months we ran Lingrail in production for a global FMCG company selling in 27 markets. Marketing and product content, May 2025 to May 2026. At the end of the year there were two invoices to compare.

Where the money used to go

Per-word rates are the visible part of a translation budget. The invisible part is everything around them: minimum fees per request, project management, revision rounds, and the weeks of waiting that push launches back.

What changed

  • Turnaround per request dropped from about two weeks to a few minutes.
  • Brand terminology stopped drifting between markets because it is defined once.
  • The team stopped managing spreadsheets and e-mail threads with vendors.

What it cost

The AI inference cost for the entire year was in the hundreds of dollars. Implementation and running the dedicated instance come on top, which is why the calculator on the home page models the full picture rather than inference alone.