Case Study

An Ongoing UI, Label, and Manual Localization Program for a Confidential Medical Device

  • Content Type: .JSON, FrameMaker, XML
  • Languages: 34 languages
  • Scope: ~40k word UI, label and ~500k word IFU localization
  • Stakeholders: writers, technical communication managers, linguists, engineers, DTPers, QA analysts, program directors, and sales, etc.
  • My Role: project Manager managing the end-to-end program and ensuring on-time, on-budget delivery with uncompromising quality
  • Tools Used: SDL Trados Studio, MT, QA software, File Compare Tool, Muitl-Term Termbase, Notepad++, ChangeTracker, Project Management Platform

Challenges and Solutions

Challenges due to lack of information:

The client provided limited to zero UI context, product images or detailed descriptions, while requesting rush delivery.

    Solutions
  • Conducting psedotranslation and machine translation to test .JSON files
  • Requesting linguists to prioritize delivery without excessive investigation
  • Guiding linguists (especially those with a proven track record of being responsible and curious) to submit queries while keeping the project moving
  • Creating and sharing a well-organized master query sheet with the client meanwhile pushing translation files to engineering to avoid downtime
  • Having linguists implement changes based on the client answers prior to final QA, followed by PM checks on key terms and high-risk areas before final delivery using a secondary QA tool
  • Managing ICR (in-country review) or third-party review for selected languages
  • Maintaining and Iterating a Q&A knowledgebase for client product and program reference

Challenges due to large word counts, tight timelines, and frequent change requests:

30 IFUs/manuals

    Solutions
  • Cross checking high-risk sections to resolve ambiguity, inconsistencies, and errors at early stages
  • Applying MTPEQA to files with high percentage of new words
  • Allocating multiple translators + one reviewer for each language
  • Building preliminary translation memories for pretranslating updated files
  • Using staggered delivery and change implementation
  • Performing centralized QA across languages
  • Iterating glossaries and global checklists as the program moves along
Stakeholder Management
High Interest Low Interest Preferred Communication My Solutions
Client A Deadlines, source queries, PM workflow, change requests Technical issues Email, meeting Workflow presentations showing value by task; monthly or ad-hoc status reports; structured communication and change management documentation
Client B Translation quality Technical issues Email, meeting Glossary creation and maintenance; close collaboration with QA analysts and engineers on program-specific changes; clear pros and cons presented for each review approach
Client C Low cost, quick turnaround, high quality Project-specific details Meeting Workflow presentations focused on how each task adds value to each project; practical, cost-saving solutions with trade-offs
Client D Technical issues Translation related issues/questions Email Launching technical tests immediately after project kick-off

Vendor Management

Low capacity and frequent delays due to holidays and internal changes

    Solutions
  • Writing heads-up emails to all affected local teams and their line managers
  • Reviewing capacity heatmap, creating and maintaining dynamic reports for local teams
  • Creating and sharing clear, project-specific one-page instructions instead of relying on system auto-generated guidance for local teams
  • Working with vendor relations to establish backup teams

Impact and Lessons Learned

    Impact
  • 0 errors found during client sampling checks
  • 0 technical issues found by client engineers
  • 0 missed deadlines
  • ~6% cost saving in QA
  • ~10% cost saving through MTPEQA
  • High client satisfaction, leading to an exclusive program partnership
    Lessons Learned
  • Although the client was very cost sensitve,
    some effort-driven service items should still be paid with higher rates or differet pricing model.
    Moving forward, I will proactively recommend different pricing for this scenario.