Optimizing Optimizely: How LanguageWire streamlined our translation process
Success in delivering a global ecommerce experience means putting an emphasis on localization. In other words, you need to meet your customers in the context of their own culture, their own preferences and expectations, and above all, their own language.
At Optimizely, we support high quality, scalable translation practices through partnerships with several language services providers. We have selected LanguageWire ourselves to create a multilingual experience across our own digital channels.
Let’s look at how LanguageWire works and how we use it.
The importance of translation
We know that most online shoppers who make up your potential global audience won’t buy from English language-only websites, as they’re much more likely to buy products and services if online information is in their own language. No matter how good your product, or how competitive your pricing, there’s a very high risk that you’ll simply be overlooked.
This realization is even more important at a time when most of us are working, communicating and shopping digitally, because digitally means across borders. It’s great to have access to new audiences in new markets, to be able to segment them and deliver relevant, personalized omnichannel content using a best-in-class content management system. But how is it possible to do this in a multi-lingual way, without losing speed and efficiency?
How we improved our web translation process
Episerver (prior to re-branding as Optimizely) faced this precise challenge. The episerver.com website operated in four languages: English, German, Norwegian and Swedish, with English being the master language. It was difficult to find a consistent tone of voice across these languages, and we faced many cases of awkward or just plain incorrect translations. There were different translations of common terms and even missing translations of high-value pages.
This was partly due to the amount of different translation agencies, as well as using internal translators, and there was no robust workflow for correcting and optimizing the varied output. We had no comprehensive style guide, and to add to the burden, documents were sent back and forth between us and our translators via email.
Three tools for consistent and accurate translation
We needed to address these problems, and we did so as part of a re-launch of episerver.com. We took the following steps:
- We put together detailed writing guidelines to ensure a consistent brand voice and uploaded it to our internal portal.
- We also compiled a termbase – a comprehensive glossary of our products and services, as well as industry terms. It includes both the master language terms and the approved translations of those terms, as well as context and explanation.
- As part of LanguageWire’s service, we began using a Translation Memory – a solution that automatically stores all our saved translations for future reference. This supports consistency, saves time for translators and therefore, saves money.
These three tools make translation – and validation of translations – easier, faster, consistent.
We were familiar with the termbase concept from our technical documentation, where a comprehensive and accurate glossary is vital. LanguageWire helped create a termbase for all our online content by performing a term extraction from episerver.com, identifying commonly used terms. We started out with around 350 terms, and it has organically grown since then. For each translation project, more and more terms are identified and added to the termbase.
The translation process
We selected our translation team from LanguageWire’s translator marketplace by providing candidates with sample texts to translate, along with our writing guidelines. We currently have two translators for each target language, with whom we can communicate directly to provide explanations or feedback.
As a translation agency, LanguageWire offers a range of services. We chose a feature called “post-editing of machine translation, plus validation.” This gives us a combination of fast automation and thoughtful human intervention. The initial translation is performed by LanguageWire’s proprietary machine translation solution, then post-edited by our selected translators and validated by our own staff. This is not only faster than all-human translation – it’s less costly too.
Finally, by implementing the LanguageWire Connector for episerver.com, we moved away from the cumbersome exchange of emails, and were able to send pages for translation from within the Episerver CMS.
The added benefit really streamlined the process by helping us to:
- Send pages and blocks off for translation from within the CMS
- Get it translated by LanguageWire
- Validate the translations in LanguageWire’s tool (Epi/Opti staff do this)
- Get the translated content back into episerver.com and simply publish the translated content with a single click
Managing translations through Optimizely projects
Our Projects feature enables the management of the publishing process for multiple related content items. For example, you can view web pages, landing pages and products and publish (or schedule) them simultaneously.
Because the Projects feature supports different language versions of content, you can manage the translation of content in Projects too, and schedule or publish content in multiple languages.
Streamlined at last
That’s how we transformed our global, multilingual presence in partnership with LanguageWire, not only improving our content in terms of accuracy and consistent brand voice, but doing so in ways that were simpler, faster and less expensive, too.
We address our audience in just four languages, but LanguageWire provides access to a network of over 8,000 professional translators working in almost 200 languages. Chances are, they’ll have your audience covered, too.
More from this series
One of our core values at Optimizely is to never stop improving and in this series of articles, we share how we are constantly optimizing our own business processes. Here are more articles from this series: