We’re at the TAUS User Conference in Seattle. One of this morning’s sessions is about translation and customer support. Panelists include representatives from leading companies, including Intel.

More and more companies are finding that they need to provide customer support in many languages, either because their product has international reach, or they serve a region where multiple languages are spoken. Customer support takes many forms, and we’ll be publishing a tutorial about how to build a multi-lingual support department.

Customer support, in most organizations, entails many points of interaction with the customers, including:

• Your website, if its well translated and comprehensive, this cuts down on the need for support inquires.
• Support forums, knowledge bases and message boards. The idea of customer powered support, championed by companies like GetSatisfaction, is increasingly popular.
• Email and support tickets. Email and services like Zendesk enable you to queue support requests, much like a call center, but without the expectation of an instant answer.
• Traditional telephone support. The most expensive form of support, but essential in some businesses such as travel.

One of the topics in today’s panel is the use of machine translation, and less than perfect translation. In many companies, the volume of material to be translated, potentially to many languages, is huge. Translating it via professional translators would both take too long (the content is constantly changing), and cost prohibitive. The consensus seems to be that using machine translation, but with places for customers to comment and discuss in their own language, is a compromise approach that enables the customers to take machine translations and expand on or correct them as needed.

Email and support ticket translation is another example. We recently profiled a company that does this, PostLingo. An optimal approach here is to machine translate incoming emails to agents (they can usually understand what was said even if the translation has errors), and then professionally translate the response (to project a professional image).

This is a large topic in of itself, so we’ll be updating this post and turning it into a full tutorial on multilingual customer support, so watch for that in October.