| Locale | Time-to-locale | Gate |
|---|---|---|
| de-DE | 41 h | pass |
| fr-FR | 63 h | watch |
| ja-JP | 29 h | pass |
| ar-EG | 88 h | hold |
Proxystreambase · Eggerness
We run a compact studio for localisation leads who need Localization Performance Analytics they can defend in a standup — cycle time, accepted strings, LQA sample bias, and the cost of a late locale.
Read the Locale Scorecard Lab outline →
Flagship programme
A working bench, not a slide course. You import a real (or realistic) export, define gates that match your release train, and leave with a weekly picture your product owner can scan in under four minutes.
The lab is built around Localization Performance Analytics as a craft: which metrics lead, which lag, and which ones flatter a vendor while hiding a bottleneck in review.
Modules, fees, and limitations →How we teach
01
Word counts describe inventory. They do not describe whether French can ship on Tuesday. We train teams to pair remaining volume with time-to-locale and a quality sample that is actually random — not the files the vendor prefers to send.
02
If your engineering train uses “release candidate on Thursday”, the scorecard must speak that calendar. We avoid vanity dashboards that look busy and fail in the change-control meeting.
03
Every hold needs an owner and an expiry. Alumni scorecards include a short exception log so “we always skip Arabic this sprint” cannot hide as an unnamed tradition.
From the journal
Why hours from string freeze to accepted locale predict missed store dates better than remaining word counts.
Read the note →
A short argument against using translated volume as a proxy for release readiness.
Read the note →
How convenience samples make a language look healthier than the backlog deserves.
Read the note →Voices from the bench
The exception log from module four is the only artefact my product director still opens. We stopped treating Arabic as an unnamed delay and started dating the hold.
Marta K. · localisation ops, Manchester
The spreadsheet templates assume a Jira export I did not have. I spent a weekend reshaping CSV columns before the dashboards made sense. After that, the weekly locale review stopped being a guessing match.
Client in grocery retail, Leeds
If you already run localisation for a GB product team and need a calmer weekly picture, start with the lab outline or write to the studio.
Write to the studio