Docs content prioritization and localization plan

Context

Following internal discussions in Slack, we've refined our approach for the different phases of the localization project.

At the time of the creation of the content prioritization spreadsheet we hadn't planned the AI-only workflow that will enable GitLab's localization engineers continuing their internationalization and feature development to get the JA site ready. This AI-only workflow will also help us deliver translations that are as close as possible to production while humans will still review the content.

We're now finishing up the latest reviews of the Phase 1 files and during this Phase, Emi and the rest of the JA team flagged some translation problems (like here) we also encountered some linting errors when trying to merge the translated files to the upstream projects.

Project Overview

In this spreadsheet you can find our docs markdown files content, divided into three phases based on page view data from October to December 2024. Please note that this data is cumulative and should be considered indicative; the final file list may change based on the most recent 2025 page view data that can be found in the 2025 datatab.

As the EN content is being edited constantly, we will need to take this into consideration when we approach the launch date to ensure our translated files are in sync with production as much as possible.

Phase Breakdown:

  • Phase 1: This phase has the least amount of files but covers the most viewed pages
  • Phase 2: This phase has more files but with medium-ranged page views
  • Phase 3: This phase has the highest amount of files but with very low page view numbers

Strategy

The team has decided on a structured approach across all phases to ensure higher quality and more consistent localization results:

AI-Only Translation Workflow

We've developed two AI-powered workflows:

  1. With TM leverage: For use after TM cleanup is completed
  2. Without TM leverage: For immediate use while TM cleanup is in progress. This will be used before launching any new language to help internationalization and development.

Phase-Specific Approach

  • Phase 1: Completed with comprehensive review and terminology standardization by the end of May 2025. Files will be re-sent to capture all target updates made during the review process and to incorporate latest EN source updates since March 2025 (request creation date).
  • Phase 2 & 3: Will initially be processed through AI-only workflow without TM leverage (due to uncleaned TM state), then re-sent after TM cleanup completion to ensure proper TM leveraging and incorporation of Phase 1 styleguide improvements.
  • TM cleanup plan spreadsheet.

Note: The "Contribute" section remains out of scope for translation.

Jobs to be done

  • Create an AI-only translation request test after @mchrastek-ext 's MR is merged.
  • Finish comprehensive review of Phase 1 before GitLab can re-send Phase 1 files.
  • Re-send Phase 1 files to capture target updates, latest EN source changes, implement English anchor IDs to headers in translated content before translation occurs.
  • Complete TM cleanup according to the designed plan in the spreadsheet
  • Send Phase 2 + 3 files through the AI-only translation workflow without TM leverage
  • Send Phase 2 + 3 files through the regular AI + MT + Human workflow after TM cleanup.
Edited by Maria Jose Salmeron Ibanez