Observer meta-research | Terms of use screen reader | Taxonomy card sort | Experience research | Eye tracking research
Towson University | Help desk taxonomy card sort
Objective: Reorganize the taxonomy for the classification of issue types in help desk ticketing software used primarily by first line help desk colleagues.
My Role: This was not an assigned project, but self-directed, “10% time,” if you will. This was a solo project that I undertook with the approval of my supervisor and the help desk application manager. I had full research autonomy including choosing test type, recruiting participants, organizing all planning and scheduling, and compiling and presenting results.
Project background
TU’s technology help center classifies reported issues in a SAS ticketing system. The data collected helps discover trends and can help direct resources. Unfortunately, the classification list had been modified over time in a piecemeal way. Staff complained that the list was confusing, there was overlap, and that it was difficult to even find some classifications. It was time for a holistic rework. I love getting my hands into research, so even though it was not part of my main duties at Towson, I was able to convince my manager of this need and secure time for this project.
Research planning
I wanted to hear what the system users thought of the classifications, which lent itself to interviewing. Finding the best way to reorganize a classifications system lends itself to card sorting. I also felt a team discussion was important to help with buy-in, especially where one or two had a different opinion than the majority.
Thankfully, there was such buy-in for this project that several main users were granted time to take part in individual and group discussions and in multiple rounds of a modified card sort exercise. The project was shaping up, with a clear need, an obvious and willing participant pool, and research methods set.
First modified card sort
The interviews helped define areas everyone agreed needed modification, for example, removing similar classifications and making it easier to find classifications.
The ticket system software requires a single list here, no nesting. For the first card sort, instead of a true sort, I asked participants to individually make comments on the current classifications. I asked them to consider renames, retires, and regroupings that made the most sense to them. I then compiled data in a spreadsheet to offer insight for the first iteration of a new list.
Things shape up
I created a new list of issue categories after completing a round of interviews, and a first round of card sorting. I wanted to bring it back to the users, so to reduce my time commitment and to foster some collaboration, we had a group meeting where I laid out best practices and presented a new list. After envisioning what a new list of categories might look like, everyone had some time to discuss questions about any changes.
Participants had a chance to offer any additional suggestions in the second round of modified card sort. I tracked each rename, classification retirement, and additional classification, to make it easier on the application manager who would have to implement these modifications.
Recommended changes
During a final group discussion, I presented the list and major modifications. I was surprised that we reduced the new list only by four items - I expected more substantial reduction.
More important than my prediction was a holistic update that included additions, retirements, renames, and regroupings. With changes that were data-driven and consensus-driven, the final presented list was enthusiastically accepted by the team.