The design team at Sumo Logic has a once-monthly ritual of a swarm. We block our calendars for a few days (warding off the product managers!), hide ourselves in a coworking space and tackle one or two large problems. The output from our swarming sessions describes a specific vision and lays groundwork for future quarters.
For a recent swarm, Rohan and I created new Sumo Logic user personas. A user persona is a fictional character who represents the actions and behaviors of a specific type of user. Personas allow designers to “create different designs for different kinds of people . . . and to design for a specific somebody, rather than a generic everybody.” Personas also aid our product managers while writing epics and the stakeholders when weighing product decisions (feature A for Melinda vs. feature B for Andre).
In creating these personas, we compiled three different types of data: what users say they do, what users actually do, and what users think about the product.
What Users Say They Do
Sumo Logic has a group of design partners who enjoy an advanced view of upcoming features and providing us with honest feedback. These design partners have provided commentary during many user testing sessions, and in ad-hoc ways as well. We contacted several of these design partners and arranged to visit their offices for a few hours for in-person interviews. These interviews delved into their daily work routines, and how they use Sumo Logic.
What Users Actually Do
As a log analytics platform, we have tons of data on what users are actually doing when they’re in our product, from specific searches to clickstreams. This gives us an idea of where users get stuck, what they do regularly and which features they never use.
What Users Think about the Product
We’ve been soliciting Net Promoter Score data and comments from many of our users for about nine months. Our Customer Success team also interacts with many customers via intercom.io, and all of these conversations are saved.
Once we collected all of this data, we combed through the commentary data and wrote the relevant quotations on post-it notes. We grouped the post-its by theme, like “troubleshooting” or “monitoring,” both common activities for our customers.
We also began using a new tool, NomNom, which provides sentiment analysis for the reams of data that we receive. We integrated it with several of our feedback collection tools, namely intercom.io and promoter.io. With NomNom, the user creates tags for specific terms that appear in comments, and then the user analyzes the sentiment behind all comments with that tag. Here’s an example of a NomNom graph, on the sentiment around “learning curve” and “query.” As you can see, there’s some negative sentiment associated with those two. (We’ll be addressing that in an upcoming swarm.)
Then we discussed it all, rearranging the post-its, running new searches in NomNom and reframing our thoughts. Three personas emerged from all of this discussion: Melinda, Kathy and Andre.
We iterated and finessed these personas within the UX group, and then brought them in front of the engineering and product groups for feedback. One of the most contentious items for feedback was the photography. Most stock photography is too polished (suits and ties) and doesn’t fit the casualness of a start-up (hoodies for days). Further, race and gender were controversial. We asked ourselves if we were reflecting the current tech industry, or if we were being aspirational about how we’d like the tech industry to look. After a few iterations, these are the current personas.
Beyond the descriptions and roles, we also mapped each persona across a few different axes. We chose the axes based on critical attributes of Melinda and Andre and how those influence their interactions with our product. For example, Melinda is likely to be calmer than Andre while they’re troubleshooting, and thus she is likely more receptive to cheeky copy in our product.
We are now presenting these personas company-wide and will update as we meet with customers in 2017.