The formatting bottleneck
Everyone in this industry has an opinion about translation quality. Almost nobody talks about what happens to the file afterwards — which is where the schedule actually goes.
The recurring costs, in the order they usually bite
Text expansion
Most languages run longer than English — routinely 30% more, sometimes far more. Every fixed layout has to be reflowed, and every reflow is manual.
Right-to-left rendering
Arabic, Hebrew, Farsi and Urdu are cursive: letters change shape depending on what they connect to. Prepared wrong, the text renders as disconnected forms running the wrong way — and it will pass an unaware reviewer, because it still looks like text.
Style drift
Word documents arrive with broken styles, orphaned tracked changes, tables that shatter on export and text boxes that do not flow. Someone fixes each one by hand.
Repetition at scale
Fifty near-identical forms across fifty jurisdictions in nine languages. The same twelve corrections, four hundred and fifty times.
Unpredictability
This is the one production managers feel most. A job that usually takes fifteen minutes occasionally takes thirty-five, depending on how mangled the source file is — and you cannot staff a schedule against that.
What I automate
- Pre-translation file preparation — cleaning and normalising source documents so they come back from translation usable rather than broken
- Post-translation cleanup — reapplying structure, styles and layout across every language version in one pass
- Batch operations — the same correction across hundreds of files, applied once
- Template compliance — documents that must match a regulatory or corporate template exactly, checked and corrected automatically
- Pre-flight checking — catching failures before they reach a client, not after
The result
On a body of recurring, template-driven multilingual work, the automation I built cut average formatting time by 63% — from roughly seventeen minutes a job to six.
The larger gain was predictability. Before, an individual job could run to thirty-five minutes depending on the state of the source file. Afterwards the average was six. A production manager can staff six minutes. Nobody can staff somewhere between fifteen and thirty-five.
The client cannot be identified for confidentiality reasons. The work received an innovation award.
How this works as an engagement
Look at one workflow
Pick your most repetitive multilingual job type. I will tell you what is automatable, what is not, and roughly what it is worth. Short, scoped, and you keep the analysis either way.
Build the tooling
Macros and scripts built around your actual files and your actual process, documented so your team can run them without me.
Run the production
Or I do the work itself — automation included — as a supplier.
