We started looking at system logs as part of our user research process while I was Director of UX at Mitchell International in the early 2000’s (gee – that still feels weird to state). As usual, necessity was the mother of invention. We needed to find out about a particular user groups’ behavior for certain roles in the system for key clients that used our solution. We needed this information to inform our workflow design for our next generation solution. We specifically needed to know how these users were using the system today across clients so we could design a better solution to better meet all our customers’ needs.
We weren’t going to trust survey methods and didn’t have the time or resources to conduct extensive interviews. So reviewing the system logs were our best way to “observe” our end-users interacting with our product. The best part is that the data is completely impartial. We didn’t have to factor in human memory or perception biases. The data told us the “story” completely unadulterated. From analyzing the system log data, we could “see” how different user types (roles in the system) across key clients were “doing their jobs” – completed their tasks. We could then conduct task analysis and define user workflow types that we could ensure to accommodate in our user experience design for our next release.