EN
Back to the archive

The archive · Work & Ways of Doing · Operational decision · 2007–2010

Microsoft's Language Quality Game: 4,500 staff review 500,000 Windows 7 screens

Microsoft turned Windows 7 translation QA into a voluntary game — 4,500 staff reviewed over 500,000 UI screens.

Microsoft

The ideaMake translation QA a voluntary game: employees review Windows UI screenshots in their native language, flag errors for points and leaderboards.substantial

What it had to solve

Windows 7 shipped in about 100 languages and rested on roughly 900,000 UI strings; machine translation and a small localization team couldn't guarantee quality, while thousands of multilingual Microsoft employees sat on unused native-language knowledge.

How it works

Before Windows 7 shipped in 2009, Microsoft faced a quality problem its localization team couldn't solve alone: the OS carried roughly 900,000 UI strings across about 100 languages, much of it machine-translated. But the company had a resource it wasn't using — thousands of multilingual employees whose native-language knowledge was free during lunch breaks.

So Microsoft built the Language Quality Game: an internal browser game where employees pick a language, review real product screenshots, and judge each translation, flagging problems for points. Moderators triaged every submission, confirmed real bugs, and players climbed personal, team and office leaderboards — participation was voluntary, with no bonuses promised.

The game pulled in 4,500 employees who reviewed more than 500,000 UI screens, including people far outside testing and localization — a French-speaking engineer could make Windows better for friends back home. It later ran for Office Communicator and Exchange, and Microsoft turned the approach into a family of productivity games, making LQG a reference case for gamifying work.

Why it lands

  • Voluntary plus fun recruited people QA could never hire — 4,500 volunteers across 100+ languages.
  • Native-language knowledge let any employee judge translation quality instantly, with no training needed.
  • Points and leaderboards made progress visible and created office-level competition (Japan took a day off to win).
  • Moderator triage kept quality: player flags were confirmed as bugs before being filed.

What it did

4,500 employees took part and reviewed more than 500,000 UI screens, at no extra headcount; the game was later rolled out for Office Communicator and Exchange, and Microsoft built a family of 'productivity games' such as Ribbon Hero for Office.

Their siteMicrosoft's own LQG guide (player & moderator roles)

What you can take

Hidden capacity is everywhere: frame routine QA as a voluntary game scored by native knowledge, and thousands of employees will do the work a small team or a machine can't.

Since then

LQG was one of several 'productivity games' Microsoft ran — including Page Hunt and Ribbon Hero for Office — and its numbers (4,500 players, 500,000+ screens) are still retold today as a benchmark for gamifying internal work. The technique worked best as a finite, high-volume mobilization of untapped employee skill, and it remains a reference point whenever companies wonder how to get thousands of people to voluntarily check something machines can't.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe the brief you are staring at, and see who has been given the same problem.

Free account · 3 free questions · no card

Related cases