We're always looking at new ways to help customers. One of the ways that have come up recently is through automated answering services. These are systems that ask you to type in your question, and then based on certain trained vocabulary, the system presents common answers to match that vocabulary.
For example, if you're calling the phone company, they might ask you "What are you calling about?" You can say "My bill is wrong." The system doesn't know anything about words like "my" or "wrong." It does understand "bill" to mean that you have an issue (positive or negative) with billing, and the system routes you to the billing department.
Systems like this are used on the web too. A certain amount of keywords (and variations like misspellings, etc) are mapped to articles for issues that the company is already aware.
Can this system work at a company as complex as Microsoft? As always, the answer is: it depends. :-)
For some of our products, there are a pretty defined set of solutions. XBOX is a good example. There are only so many things that a customer can complain about with the XBOX - hardware, software, the XBOX Live Service. Each of those things have a pretty defined set of failure points that can be well documented and mapped.
This is a really hard thing to do for open-ended questions though. How can you train an automated answering system to know what to do with a problem like "Every time I try to replicate Active Directory across my two trees, it takes about 4.5 hours even though I have the sync interval set to every 4.5 seconds." As you can tell I'm not that technical, but you get my point. It's really hard to take a generic question like that and present anything better than maybe the most popular Active Direction replication articles.
If you're going to get that granular with your issue, you're better off with a good search mechanism rather than training the system through automated answering. The issue then becomes harder for every piece of content you add to the system, and the traditional search issues begin to emerge.
Are you in a technical company trying an automated answering system for customer issues? Is it working for you? Share your stories in the comments of this entry.
Hunter DonaldProgram Manager - Knowledge Management StrategyCommercial Technical Support