“What do you see as the biggest challenges in creating a global-scale customer support environment today” A Q&A with Robbie Liotta of Dassault Systems transcribed from SDL’s recent Localization Roundtable in Boston.

Dassault specializes in mechanical CAD content and I focus on the technical support strategy. A couple of years ago we started to review the global support strategy – we found there was a huge library of support questions that were hosted in customer accessible FAQs but only for English speaking customers. Support requirements for non-English speaking clients had grown significantly and we saw no reason for that growth to slow – but non-English support was all via expensive call centers staffed with personnel who were multi-lingual. We set ourselves a goal of call-avoidance for redundant topics with the objectives of saving costs associated with 5-6 support personnel and improving the non-English speaking customer experience.

 

We began by tracking which English entries were hit most and selected any that were hit five times in a certain time period. We then selected languages – French, Spanish, both Chinese variants, Italian and Russian while keeping a watchful eye on other languages like Portuguese as markets like Brazil generated needs.

 

What we noticed as we approached this was that the English answers were often long and complex – not written in a style that made them easy to translate. Making the authors aware of what we needed – both in terms of writing style but also in terms of repeating common phrases (that had been previously translated) were also important. 

 

Once we knew what we wanted to do we had to confront how to achieve it. For the translation we reviewed all our options; clearly machine translation wouldn’t provide the accuracy our clients needed but traditional human translation was inordinately expensive and time consuming – especially given the high word count. SDL approached us with their Knowledge-based Translation System – a highly leveraged machine translation environment coupled with human post-editors. KbTS is offered as a service to clients with guaranteed service levels – but they told us the system only applied to certain types of content.

 

When we balanced the costs of continuing in the current fashion versus the benefits it seemed a no-brainer and our experiences to date seem to have born that out. Once you exceed a million words or so the technology really justifies itself because it builds on the other linguistic assets, adds the trained machine translation but retains the human touch to provide the perfect quality we need. 

 

 

 

 

 Question: What’s the cost difference in localization you experienced?

Answer: There are three levels to consider, pure machine translation, the KbT service and fully human translation, obviously backed by TM – in theis case SDL’s Trados. Pure machine translation is pretty inexpensive but not appropriate for what we are trying to do at the moment but we are looking at how we can use it in the future. Knowledge-based Translation is provided as a service rather than sold as a standalone product. The service guarantees quality and timeliness for clients who have appropriate content and although I can’t remember the word rates it’s something like two thirds of the cost of traditional human translation.

 

 

Question: What kind of training did you have to give your support engineers who are doing the tech writing?

Answer: Our situation might be unique because we have such complex software products and a significant amount of industry terminology. The whole process was pretty interesting though and training our authors was central since we had to make sure they used language consistently – especially the specialized words commonly used in our industry. As technical support staff, writing isn’t exactly our core competency so we did have to create training courses for them to cover basic lessons on how to write in a way that would be easier and less expensive to translate.

 

Lessons learned.  

To make this viable the benefits have to outweigh the costs. The basic points to make sure of: 

 

1.       Quantify the things to be translated using some method – we track how the English language FAQ entries were used and when the count exceeds five we have it translated.

2.       Pick your languages carefully based on the markets where support costs are highest

3.       Make sure the source content is simple and easy to translate. Train your writers so they can support the global support efforts.

4.       Make sure you have translation technology to support your needs. For us the KbT was appropriate, especially once we had the authors using the Author Assistant tool to make the English answers more consistent with style guides and previously translated content.

 

Robbie has offered to answer any questions you have. Post them as comments and he will reply.