When I talk to other companies about knowledge management, I realize that Microsoft has a fairly unique problem - we really care about providing information in a language you can understand. This is a really neat problem to have though, since as a corporation, we care about a wide variety of customers and recognize that as US-centric we are, we really want to make customers realize their full potential with our software.

There are several downside to enabling such a wide variety of languages for our support content for customers. Here is an idea of some of those issues. If your organization is thinking of expanding the number of languages you support in your documentation, I hope you find these tips help frame your future discussions. This list isn't exhaustive of course - more of just a brain dump this morning. :-)

  • You have to write everything in a "global English" voice.
    • This is really hard if you want to empower everyone to write content. It's hard to train my organization's thousands of support people how to write in a voice that machine or manual translators can understand.
  • Machine translation isn't very good.
    • There are some languages that do better than others, but that assumes that the community consuming the content is tolerant of machine translation. If you want to rely on computer-assisted translation, check with your customers to see how they are willing to deal with the inaccuracies that machine translation provides.
  • Translation is still really expensive to do, even with automation.
    • You'll still need an army of editors to correct mistakes that customers find, especially with high traffic articles.
  • Consider creating a tier system for your content.
    • It's unlikely that everything your organization writes needs to be localized into the languages you want to support. You should have a hard conversation and really think about which content needs to be localized at all, and then a deeper discussion about how to do it.
    • There are three tiers that I think of:
      • Person translation - highly complex, mix of buzzwords and speech content. Also good for your most popular content in the regions you're translating for. The stuff people see the most is what you should be paying the most attention to.
      • Machine translation - good for stuff that customer's will look at occasionally. If you start to get a lot of complaints on the poor translation of several articles, you can always go back and have a person clean up the translation (potentially for a reduced cost).
      • Don't translate - I think this one is something we have to ask ourselves at Microsoft more. In the IT Pro and Developer customer space, we find that many of our customers are able to read English fairly fluently. Their tolerance for English content is higher than a consumer's tolerance.

When we increased our blogging efforts from our technical support groups, we had to determine the best way to ensure that all customers would be able to view our content. Part of that discussion focused on localization. After a lot of discussion, we decided that because our customer base was tolerant of English text, and because we do have services like the Windows Live Translator, that should fill our content gap enough to satisfy customer's demands. We did talk about formally translating some blog entries, but we realized that the balance of work required wasn't worth the effort yet. I'm sure we'll occasionally revisit this as we continue to ramp up our blogging efforts in the next 12 months.

Is your company trying to figure out how to manage multiple languages? Drop a line in the comments with your concerns and we can start a discussion.

Hunter Donald
Program Manager - Knowledge Management Strategy
Commercial Technical Support