Dell heads up Booking.com’s 400-person, in-house content agency. While the department handles editorial, moderation, marketing, copy, and geodata tasks, among other work, about 250 of its members are focused on localization.
More than 100 groups of stakeholders regularly need to translate and localize a vast range of content for desktops, browsers, and, increasingly, apps. Collaborators include brand marketing teams, PR, legal departments, and UX writers.
The scale of requests is massive, on the order of tens of millions and hundreds of billions, depending on the task. Dell explained that the team has designed approximately 20 services or solutions that can be applied to bundles of requests, depending on the metrics that will be used to measure the success of a particular type of content.
Booking.com has a three-prong approach to localizing the platform, which starts with localization professionals, both in-house and, as needed, third-party. In many cases, these experts have more than 15 years’ experience, as well as a “very good feel for the brand culture of the company.”
Despite the “significant overhead, in terms of management,” Dell said these personnel bring three major benefits: quick turnaround; a deep understanding of technically complex products and their corresponding terminology; and local insights.
Machine Translation Pays Off
Booking.com’s own machine translation (MT), the second pillar of the localization program, provides the majority of MT output the company uses.
Dell started up with MT six to seven years ago, when significant interest among Booking.com’s large machine learning community made it easy to recruit volunteers to the project. Since then, freelance translators have helped build Booking.com’s very large, clean, high-quality corpora — work that continues to pay off in some unexpected ways.
“Some of the tech that we built around machine translation […] we also started using it for moderation and things like that,” Dell said.
Slator Machine Translation Expert-in-the-Loop Report
60-page report on the interaction between human experts and AI in translation production, including AI-enabled workflows, adoption rates, postediting, pricing models.
The final prong in Booking.com’s localization model is creating content in a non-English language, a solution that, while more appropriate for some markets and content types than others, is becoming more practical as AI continues to make strides.
Dell said his team has “already crossed paths with some ChatGPT-like products,” for MT as well as automatic classification or taxonomy of content, and that experimenting with new tech is “definitely baked into our kind of thinking.”
Ideally, he added, technology will advance to help quality management work around or even avoid less-than-great source material: “My dream, ultimately, is to generate content in many, many languages for many, many different kinds of use cases.”
For those who missed SlatorCon Remote March 2023 in real-time, recordings will be available in due course via our Pro and Enterprise plans.