All Posts
Roughly a year and a half after the launch of ChatGPT, AI remains a hot topic in the world of tech, and for designers. After trying out a variety of AI design tools, it has proven to be helpful in some areas more than others: during iterative phases, to organize digital workspaces, or to write copy. While we can and should be using it to brainstorm and save time with busywork, we can confidently say that it won’t be taking designer’s jobs anytime soon. Any AI design software we’ve used that has tried to complete a task requiring critical design thinking has produced results that range from generically mediocre to outright comedic.
Several new features were announced at Config 2023 for both Figma and FigJam, some of which utilized AI. One that we were curious about was FigJam’s Sort stickies, specifically as a tool to sort user interview results during the discovery phase of a project. We were curious about it because it claimed to do exactly what we had performed so many times: to look at a ton of info on a ton of sticky notes and group them based on similarities. While some designers don’t enjoy this phase of a project, our designers find it very satisfying to find common themes in a mess of information, and to make sense of what seems like nonsense. But we will admit, it can be tiring, and time consuming.
These days, when we do user interviews at Crowdlinker, they tend to be with 3-10 people, depending on the project. This produces a pretty reasonable amount of data to sort through manually. In the past, we’ve done co-creation workshops with groups anywhere from 10 to 80 people. As you can imagine, the more data there is to be sorted, the more useful this type of tool becomes.
To test the Sort Stickies, we had the results from four user interviews on stickies for a project that was healthcare-related. Our designer and product manager had conducted 2:1 interviews with healthcare professionals and asked about 10 questions per interview, each interviewee’s answer being placed on a sticky note. Our designer took the results and sorted them manually first, to see what themes we would draw before being influenced by the tool’s suggestions. We then got Figjam to sort them.
So how did it do?
It came up with nearly exactly the same themes that we had, which was both exciting and a little disappointing. Exciting because it nearly replicated our work, disappointing because (maybe expectations were a little too high) we hoped it might identify themes we hadn’t. We repeated the process a couple of times, and each time, the results were different, sometimes more useful than others.
It became clear quickly that how things were phrased on the stickies had a huge impact on how they would be grouped. For example, one of the stickies simply said “Personal computer for the clinic where he works.” It didn’t have any context, and it didn’t capture whether using the computer was a preference or a rule to be complied with. Considering this, we would have been thinking more about how we were phrasing both the questions and capturing answers if we knew we would be using this tool to sort the results.
Another interesting consideration was colour-coding stickies. All the stickies that were a reply to a specific question were coloured the same, which made it easy to see if there was a common theme in response to one question. For example, a theme that Figjam identified was “Clinic Variations,” and we could see very clearly that most of the stickies in this category were in response to a question about paperwork. A key pain point identified was the lack of standardization in paperwork processes across clinics. While it’s a small consideration, anything that can be done to ease the cognitive burden while sorting data is helpful.
Even after optimizing phrasing and utilizing colours, we were still left with a longing for something more from this tool that would help it produce more nuanced results. We would have loved to have been able to provide some prompting for FigJam to consider in the sorting process, such as sorting based on emotive qualities, specific verbs, or other details. Using this type of sorting method several times using different focuses could help create a thorough, holistic list of themes from the user interview results, and in turn, form a more thoughtful, comprehensive set of design principles.
Ultimately, is Sort stickies useful for the user interview synthesis process? My opinion is yes, and the more data you’re dealing with, the more appealing it becomes. To utilize FigJam’s AI tools further, the Summarize stickies feature may also be useful as a secondary step to write a brief summary on the themes found in the cluster of stickies. But as with all AI tools, the output needs to be reviewed and further refined by a human. If used intentionally, Sort stickies can certainly cut some busywork from user interview analysis.
