Localization at Scale
Localization at scale is the practice of treating multilingual publishing as a feedback system and a long-term knowledge system, not as a one-off translation task.
Core idea
Localization is broader than translation.
It includes:
- source writing
- terminology
- metadata
- navigation
- accessibility text
- diagrams and screenshots
- review workflow
- CI and QA
- drift management
- native-speaker review
Authoritative handbook
The central operating model for this initiative now lives in:
../localization-at-scale-program.md
That handbook defines:
- why multilingual publishing matters
- why localization is a system, not a project
- how AI fits inside feedback loops
- the quality model of coverage, completeness, quality, and freshness
- localization debt and open defect classes
- source-first remediation and AI-agent workflow
- localization drift and repository maturity
- the detailed Native-Speaker Review Findings corpus
Practical rule
A localized page is not complete until all reader-facing content is localized, including summaries, tags, body copy, captions, diagrams, alt text, and related-content surfaces.