Many thanks to James Ayscue for inspiring this topic.
We all know that support websites are important. But how do we know that they actually work?
The first step is to attract customers to the support website. Many will visit from a web search and others will go to the home page first, and either method is fine. Don’t bother counting page views: they are meaningless. Count unique visitors, and keep an eye on unique return visitors: those are the folks who are finding value in the website.
But volume does not equate to success
Do website users find what they need? That’s not easy to measure because it’s difficult to capture intent, that is, what a user wants to achieve by visiting the website.
One method is to study paths through the website. For example, if a user types a query, reads one document, and leaves, chances are that they found what they wanted, but if the path is query/read/log a case, it may point to a failure.
The other method is to simply ask users whether they found what they wanted, when they exit the site. With a one-question survey the response rate should be pretty good and since results are quite stable you only need to do it a couple of days a year.
And it does not equate to case deflection, either
Many support organizations want to prove that their website somehow “prevents” cases. It breaks my heart a bit, since we should be trying to solve customers’ problems rather than pushing them away–but if you are interested in measuring case deflection you cannot equate website visits, even “successful” visits, to cases. Many users visit the website but would never log a case instead! This is especially true if you have lots of functionality available on the website, or if you restrict the number of customers allowed to log cases.
That said, you can certainly calculate interesting metrics around case deflection, e.g.
(KB and forums users – users who log a case within 24 hours of the visit) / (KB and forums users) = % potential deflection