How to organize TRs & MRs to meet Dec deadline

Background

We have a launch date for Tech Docs in Japanese as 10 December 2025. The ultimate dream goal is to hit 100% on launch date. The realistic fluctuating goal is to meet 30-40-50% of the content translation in time for launch.

Problem statement

Figure out an efficient workflow of processing hundreds of files in translation (content being technical documentation) under a tight deadline of 6-7 weeks. We have the following variables at play:

  • Content: 1,712 files to process in 6-7 weeks
  • System for planning and tracking: GitLab
  • System for request management for translation: Argo
  • System for translation management (where linguists and PMs can work directly): Phrase

We identified the following critical steps in the workflow:

  1. an .md file is sent to translation from GitLab to Phrase via Argo
  2. an .md file is translated via AI script (Translation step)
  3. an .md file is QA'ed w/ the help of automated checks and built-in Phrase QA checks (AI QA step)
  4. an .md file gets released to GitLab in the form of MR and moves into Revision step in Phrase (to be reviewed by a human linguist)
  5. an .md file gets reviewed in GitLab's MR and merged to master to await the official launch in 6-7 weeks

Solution

Break the 1,712 files down by milestones as follows (check off items as they are completed and merged into master):

Milestone 1 - Critical & Dependent:
  • files that fall under the criterium of those currently blocking engineering tasks, e.g. those with anchor IDs missing
  • files that fall under the upgrades & deprecations, e.g. those from this issue
  • files pertaining to GitLab Duo within any directory of the tech docs repos, e.g. specified in MJ's slack message
Milestone 2 - Important & Customer Facing:
Milestone 3 - Good to have for launch:
  • Everything else without any particular priority

Cadence

1,712/6 weeks = 285 files/a week to translate, QA, merge (just AI output)

Based on our estimate, discussions w/ the vendor, engineers and stakeholders, what's realistic to target? - check off the mark:

  • 30% = 514 files
  • 40% = 685 files
  • 50% = 856 files

What to exclude

Redirect files. They will be placed under a separate component in Argo (request mgmt system) to not show up as new/updated assets on the dashboard hence reducing effort for us in translation management of tech docs.

Edited by Iryna Rogova