When troubleshooting a technical issue with Mentor tools, most customers who visit SupportNet use search. In fact, search is the most-used feature on SupportNet. Feedback collected from the SupportNet survey indicates 2/3 of the time customers find answers to their troubleshooting questions.But what about that other 1/3 of the time when the answer isn’t found?
Well it’s our job to find ways to improve these numbers and I’m happy to announce we’ve started down the path of making further improvements to SupportNet search effectiveness.As a first step, we are now using the Google Search Appliance (GSA) as the search engine on SupportNet.
For the last few years we have used a natural language search engine.When answers weren’t found, customers often said “Why can’t you just use Google?” If only it were that simple.Surprisingly, finding an answer on a technical support web site is a different problem than locating information on the Internet. An accurate technical troubleshooting answer depends on a good search algorithm along with other factors that include: content, metadata, product filters, site navigation, as well as the search query.
To determine whether to change our search engine, we ran a comparative A/B test for about 6 months in 2010 where half of the SupportNet sessions involving search were directed to our existing search engine and half of the sessions were sent to the Google Search Appliance.Then we analyzed the data to see if either search engine produced higher troubleshooting success rates or visit satisfaction, and whether there was a difference in how many people opened a service request after searching.As it turned out, we had a tie.There were small differences between the two search engines but most were not statistically significant.The significant difference we detected with Google search was the faster speed with which it returned results.
Next we will begin tuning the other factors beyond the search engine to increase the overall effectiveness of finding an answer on SupportNet.You can help us improve search effectiveness by continuing to let us know when searches don’t work.The SupportNet Survey and the Feedback link on the footer of every SupportNet page are excellent ways to share your thoughts about where we can improve.
We would also like to better understand your expectations for search. Are there certain types of problems that tend to produce good answers with SupportNet search? Are there issues where you have low confidence about finding an answer?You are welcome to post feedback below or send an email to firstname.lastname@example.org.Don’t forget to try the advanced search page for more robust options and search tips. I look forward to hearing from you.