Machine Translation and Post Editing Drive Volumes

A successful retail business can grow like gangbusters when welcomed by millions of enthusiastic customers around the world. But do current production models scale with that growth?

Our client—one of the world’s largest online retailers—had exploded on the international scene and showed no signs of slowing down. They implemented a home-grown machine translation (MT) engine and workflow to reduce workloads for human translators and allow sky-high volumes to be published.

But the raw machine translation was not hitting the required quality level for most content types.

At the same time, our client struggled to integrate and analyze quality measurement metrics so they could make strategic decisions about pricing, throughput, and resourcing.

Finally, the client’s workflow needed to be redesigned. Manual administrative and production tasks cost hundreds of hours of employee time per month.

Here’s how we did it

  • Designed and implemented a post-editing program—a human review to bring raw MT up to agreed quality levels
  • Recruited, onboarded, and trained a team of 38 highly experienced post-editors
  • Created an automated task that ran a quality scoring algorithm to determine the effort needed to turn the raw MT into high quality translations
  • Streamlined manual tasks
  • Built automations to create jobs, download source files, convert them to the appropriate format or post-editing, and deliver files after completion


  • Eliminated 200 hours of manual tasks per month
  • Allowed PMs to focus on higher-value tasks
  • Increased quality of output
  • Achieved nearly 7.5 million words of post-editing in one year
  • 30-fold increase in volumes processed in 6 months
  • Provided solid data that informed pricing, quality and resourcing