Using Internal Resources to Coordinate a Multi-Language Information Management System

Editor's Notes

• #2: The reason I wanted to come to LavaCon was to learn how to improve my translation/localization workflow by learning from the best in the field. I am not a content strategist—I am not a Manager or Supervisor. I am just a technical writer tasked with coordinating the translations of our content. Case study. I don’t proclaim this is the best way—in fact I am recognizing that it is far from perfect, but that is why I am here. I hope that this conference, and this session, will help me to bring back to my workplace information to improve
• #3: Pie chart represents Web Self-Service onlyTechnology Services Industry Association - (TSIA)I’m starting really broad and will narrow down to my example
• #4: About my company, how many languages we localize content intoHow many continents are there?Over 180 countries (doesn’t matter, will discuss later)MarketsDo I need to briefly discuss translation versus localization?
• #5: ROI STUFF HERE – we’ll talk about later, starting with the top-content (BIGGEST BANG FOR YOUR BUCK)
• #6: Markets are not the same as languages
• #7: Make this transition to the Ideal versus Real world, “Things Needed” are Ideal, “Things used” are real
• #8: LAVACON helped to push me to document the process so I could analyze itPLAN – DEVELOP (localize) – DEPLOY (is it effective)Reverse—document what we’ve done instead of following a strategySet expectations = get buy-in OWNERSHIP/MANAGEMENT AND BUY-IN FROM THOSE WITH THE RESOURCES YOU NEEDWe had 1 French speaker –but I argued that we should have centralized content retrievalStick to the workflow = trust it because if you change it, many problems can come up—then you end up with no standards and no consistency
• #9: It’s not which is best, but which can I use for my purposes?How does Your Company go to market?Same products – Centralized ModelDifferent products – Regional ModelStrategy is Real-World. We work with the users who want the content—not guess. That’s how we are able to localize.
• #10: We use Centralized model
• #11: Our goals—your goals will be different –this is why I wanted to present this as a case studySharon Burton – content is a reflection of your brand.Content strategy– you know that one strategy doesn’t work for everyone. Keep in mind my company profile and the brand perception we are trying to have
• #12: We use Centralized modelWe control KMS (and want that control) –consistency of brandThe translators may not be employees so we can give unfettered access to our KMS server
• #13: Make this transition to the Ideal versus Real world, “Things Needed” are Ideal, “Things used” are real
• #14: How is this? By choosing what kind of localization model you’re going to use, users will come to expect it. Difference between Google translate, machine translation, localization companies—if each article/script is translated using a different method, you are setting an inconsistent expectation about the quality of content. I work at a security software company so accuracy will trump speed in the content lifecycle.
• #15: Establish your localization team -establish content workflow – it’s important to have the language czar, and it’s important that the job of translating support content isn’t an annoying side task that is outside their basic job scope. In order to make it work long term and in order to ensure that content in other language is updated quickly, the alternate language team should translation as a significant part of their job description, or it won’t work.One of the conversations to have at this point is to determine if the alternate language team is given the ability to create english content as well. That’s a decision to make based on their technical prowess, but unless you have a good reason why they shouldn’t create English content, you should make sure that this is easy for them to do and that the workflow allows for it. You don’t want to limit knowledge creation to the resources you have at headquarters.2. Determine which content will be localized - begin immediately – This part of the implementation process will almost always take longest (depending on how far along you are with your implementation, you can determine whether it makes more sense to input the translations directly into the system (preferable) or you can do the tranlsations outside of the system and import them in.)3. Localize UI4. Tune Search5. Test / Soft Launch – If possible, always do a soft launch.Will take 2 months, assuming you have already launched in English. It’s possible to launch 2 languages at once of course, but you’ll obviously have to make this judgement based on the amount of resources and demand. If possible, always best to get one languagerunning smoothly so you can learn from that and improve the next time.Obviously this is a rough estimate, and is based on our experience, but we’ve launched 4 language
• #16: 85% of traffic / 15% of views
• #17: http://www.kilkku.com/blog/2012/09/the-final-obstacle-to-wiki-tech-comm-localization/We have worked really hard to do this. We have recognized that our language markets are unique and the responses/content that we send users must be to that level as well.
• #18: --Our global market share was also increasing, but we didn’t even have all the English support content that we needed, much less the alternate language content. --Finally got a huge chunk of knowledgebase content in one place, but were using an open source solution that didn’t allow for --Poor Search, limited multi-lingual support. When I say multi-lingual, I mean search, reporting, MY PRIMARY RESPONSIBILITY IS CONTENT CREATION NOT KMS ADMIN
• #19: ROI STUFF HEREZAP numbers 201 total English ZAPs (78 used last month)85 Spanish84 French
• #21: The advantage of the cumulative approach also creates a dilemma. If I receive only one request or if there are conflicting requests pertaining to the same word or phrase there must be some process used to determine which word or phrase to use. I have had success making this determination myself, as a non-Spanish speaker, by emailing the group the translation request and asking for feedback. Generally, I will receive back about 5 replies with that agent’s reasoning for using or not using the translation request. The basis for my assertion that having a coordinator who doesn’t speak the language is this—if I spoke Spanish, I would be opinion number 6 and just one more suggestion on which to decide. Instead, I can look at where each translator is located, who the primary audience for the deliverable is, and all other relevant factors; then an objective decision can be made.
• #22: Right away I can see that “contactar” is used in all other ZAP scripts, so making that change would require changing all the others, unless this is specific to this situation