Lead Product Designer responsible for Sumo Logic's metrics query and dashboard creation experience.
Sumo Logic is a cloud-native real-time machine data analytics platform that empowers teams around the world to make data-driven decisions.
Let’s break it down further. Assume that you work at a large airline company and encounter co-workers named Andre and Kathy. Both of them use Sumo Logic but they have different usecases in mind.
As a Software Engineer on the Infrastructure Team, Andre's usecase involves making sure that the systems are up and running as soon as possible whenever the airline's check-in infrastructure goes down.
He uses Sumo to monitor the health of the server infrastructure, to get notified of system failures, and to troublshoot the root cause of problems and fix them quickly.
On the other hand, Kathy who heads the Business Operation team is responsible for expanding operations in certain regions of the world by analyzing airline's passenger traffic and travel patterns.
She uses Sumo to create dashboards that give her rich insights about her team's target for the quarter, understand popular routes and trends, build strategy and product roadmaps along with key stakeholders and business partners.
While both these users use Sumo Logic to solve their respective problems, there is a problem. Users are lost trying to figure out how to write a query.
We followed the Double-Diamond process to gather research insights and uncover user needs, scoped the problem at hand keeping in mind the product release timelines, designed prototypes of varying fidelities, tested them with our users and delivered designs throughout the solution phase. This project was possible due to the close collaboration of teams across the world (we hopped on calls with teams in Poland and India).
As the lead designer on this project, I scoped out the specifics of the problem by collaborating with UX Research, Product Management and Engineering.
Based on qualitative data from our user interview research sessions and Nielsen-Norman's Heuristic Principles, we were able to identify 4 major themes.
Both Kathy and Andre found it hard to write metric queries during critical times. By understanding the semantics of the query language and observing how users grouped data, we broke down a query into four broad categories: Metric, Filter, Aggregation, Group By.
Neither Kathy nor Andre knew what to start typing. While they wanted some level of guidance, they also wanted the flexibility to gather unique insights. Through user interviews and query writing observations, we realised the need for an autocomplete component.
While Andre wanted to monitor the server health of us-east and us-west regions, Kathy was interested in popular travel places/times from San Francisco and New York City. Deciding the level of freedom that users should have when using a structured query builder was the core question.
After multiple iterations, based on user feedback, implementation constraints and timelines, the degree of freedom was decided. Two menus would be shown depending on whether a user is just starting off with the search for either a key/value or picking a value after a key is selected.
With the introduction of the structured query builder, there were key concerns that had to be addressed - "How do we switch to the other one? Does this tool handle all kinds of queries? I want to be able to modify queries using the query language too!"
While this was an essential function, the users neither wanted it to stand out explicitly or be hidden within menus. This led to finer interaction design questions.
Andre who was an expert user pointed out nuances and edge-cases that had to be addressed. For instance, Andre often leveraged multiple complex queries to get to the root cause of his bug. He wrote a second query to take in results from the first and feed it to the third.
Error handling and accessibility were other factors that we looked into at this point.
Both Andre and Kathy were able to easily use the new query builder to gather insights about their respective problems and solve those faster. This new solution reduced the query writing time by half and improved satisfaction scores, thereby leading to happy users.
This product was released as a part of the larger Kubernetes initiative at the company's user conference.
With the help of the new solution, Kathy was able to successfully create dashboards that informed her product strategy using data about trending tourism spots, social media tags and her airline's passenger traffic throughout the year. On the other hand, Andre was able to single-handedly pinpoint what led to the system breakdown and get the systems up and running without having to alert another soul.
We used HEART Metrics to measure the success of this product and we found that users were extremely happy that they needn't learn a new query language and could complete writing queries a lot faster.
