Chapter 2
Why the legacy translation model breaks at scale
Existing translation workflows were built for static, predictable content, and even with modern technology they're still guided by humans.
It’s people that own the process, and the quality of the translation is down to them. In return for accuracy, they require time and budget. The greater the risk of inaccuracies, the greater the time and budget required.
But in an era of high-volume, high-velocity content, these trade-offs are no longer an option. Enterprises can’t accept that accuracy and control must be paid for in weeks, with escalating costs for every adjustment and manual amend.
Machine-assisted translation surged 533% in 2024, yet workflows remain largely human run.
Source: Forrester Research, 2025
Share of enterprises still reliant on fully manual translation workflows
JP: 0%
US: 0%
DE: 0%
UK: 0%
FR: 0%
Source: DeepL 2025 survey of global business executives
“We all knew language was a barrier and challenge, but I don't think we realized how much until we started using DeepL Voice and seeing the impact that fully understanding what someone was saying in their native language really had on us.”
Jodi Sweed, VP Strategy & Development, Aramark
The rise of voice translation is a powerful example of how translation processes re-imagined with AI are becoming essential for meeting new expectations.
In our survey, almost two-thirds of business leaders describe real-time voice translation as significant to their operations. When asked what’s driving this, they point to the key factors listed below.
Behind the shift to voice translation
0%
cite advances in the accuracy, speed, and ease of integration
0%
see rising customer demand for real-time communication
0%
highlight the pressures of expanding into global markets
Source: DeepL 2025 survey of global business executives
