AnthonyHughes
DragonPass 2026 Case study

Quick Localise: shipping design in 22 languages without the busywork

A custom Figma plugin and merge pipeline that turns one master design and a translation spreadsheet into 22 review-ready localised comps.

22languages automated
1plugin run replaces a manual slog
0reconciliation steps
Quick Localise: shipping design in 22 languages without the busywork — project imagery (placeholder)

Problem

Mastercard Travel Pass communications ship in 22 languages. Translations arrive as a single Excel file — one column per language — but producing localised Figma comps for stakeholder review meant manually duplicating frames and pasting strings, language by language, template by template. Slow, soul-destroying, and error-prone: exactly the kind of work where a missed string or wrong-column paste ships to a real market.

Objective

Automate the path from “one master design + one translation file” to “22 review-ready localised comps” — removing manual reconciliation entirely.

What I did

  • Evaluated buy before build. Assessed the Figma localisation tooling landscape and identified the recurring friction: the translators' file structure never matches the tools' expected import format, forcing manual reconciliation across 22 columns every single time.
  • Built the merge pipeline. A Python script that auto-detects the source-language column, maps locale columns by name and ISO aliases, joins the translators' strings onto the import template, and outputs a QA report flagging unmatched rows and unused translations.
  • Then built the plugin itself. Quick Localise — a custom Figma plugin that reads the translation file directly, clones the master frame once per locale, loads fonts, swaps text by layer name or source string, applies RTL alignment for Arabic and Hebrew, and reports missing translations and unmatched layers. No reconciliation step at all.
  • Made an honest build-vs-buy call. Documented exactly where the custom plugin is deliberately thinner than commercial tools — engineering judgement about where polish is worth paying for versus owning the workflow.
  • Solved the typography layer too. Researched Mastercard's multilingual brand typography, including script coverage gaps, to define script-by-script font handling for the localised templates.

Outcome

Localised comp production went from a manual, per-language slog to a single plugin run — with built-in QA reporting catching missing or orphaned strings before stakeholders ever see a comp.