Users wanted to analyze data without knowing SQL, so we provided support for interactive analysis that became a highly adopted feature.
Why Did We Built It? Who Did We Help?
Periscope Data helps data teams to answer critical questions. Increasingly, that means democratizing data in the organizations and expanding the access to everyone.
This project was a great opportunity to empower broader audiences, such as salespeople, marketers and other business people, and save time for data analysts by operationalizing their work.
Data analysts have upper limit on building dashboards and ad-hoc analysis
By enabling self-service for partner teams, the data teams would gain more time to focus on advanced analysis.
Business Users are forced out of the product to answer simple questions
A lot of users would download CSV and conduct analysis in Excel or would use another, competing tool, given that they were not able to perform an analysis without SQL knowledge.
Timeline
December 2017—May 2018
Team
UX Designer, PM, Engineering Lead
Uncovering Business Users' Stories:
Prior to kicking off the project, I have organized remote user interviews to understand our users' motivations, desires, needs and context. Our team learned about business users' collaboration with data teams and IT departments, their comfort of joining data, as well as their roles and responsibilities. These findings provided input for the product manager while defining project requirements.
As a business analyst I want to play with relevant data to discover trends.
As a business analyst I want to build a dashboard for my team.
As a data analyst I want to be able to audit an interactive chart.
A lot of the users had a clear mental model of drag and drop interaction, hence we decided to focus on exploring and testing with users the proposed interactions and workflows. Before diving into wireframes, I researched BI tools to better understand the industry standards and workflows.
Wireframes, Testing and Visual Design
After defining the workflows and requirements through rounds of user interviews, stakeholder interviews and engineering reviews, I created wireframes for usability testing.
As a part of the usability testing I explored the following:
1) How our users think about analysis creation
2) Auditing an interactive chart
3) Interacting with charts of mixed origin
4) Aggregation, filtering and limits
Once we confirmed direction and reached stakeholder approval, I started working on the visual element of the experience—looking at leveraging our current visual language and expanding it where appropriate.
Testing calculations.
Exploring interactions, layout and visual design.
Interactions
The central interaction of this project was to allow users to create code-free analysis. I explored three input types: dragging and dropping into designated fields, typing into fields and dropping a field on the X & Y axis of a chart. The last method was by far the most popular, however, it also proved to be the most challenging engineering-wise. We opted to add it to the product in a future release.
Support keyboard accessibility by allowing users to add fields by typing the values within the form.
Easily manipulate table chart type by re-arranging the order of the fields.
Provide business users with context via Dataset tooltip.
Internal testing and release
We released the feature internally and gathered feedback through a dedicated slack channel. Additionally, we organized a contest for Periscopers to created the best dashboard.
Later in the process, I collaborated with engineers to make sure that the working solution is following the design and we were addressing any edge cases we might have missed.
After releasing the feature—called Data Discovery—to our users, the PM and I joined feedback calls in order to identify areas of improvement.
Constraints
Due to the requirement that business users' experience was to be accessible to data analysts, a lot of the layout decisions were informed by the editor layout. I explored alternative ways of entering the interactive editor, but ultimately our users gravitated towards an existing flow of creating a new chart from a dashboard.
Since data analysts needed to be able to create and interact with both editor types, we opted for a toggle on top of the schema tab for controlling the editor experience.
Data analyst switching between SQL and Discovery editors.
Business Results
Most of our high valued accounts have had Data Discovery enabled and over 40% of those have already incorporated the new workflows in the first 6 months. The feature has also been an attractive driver to our product, with many of our sales reps mentioning that Data Discovery have attributed to their successful closes—both renewals and new businesses.
Lessons and Challenges
Challenge—Discoverability
Our biggest challenge turned out to be feature discoverability. Since we prioritized data analysts' experience in the product—getting as quickly as possible to query building— we compromised feature visibility. We planned to address this during our work on Information Architecture and further enhancements to business analysts' experience.
Lessons—Work with an area expert
The PM assigned to the project had a deep understanding of the audience based on her background in analytics. Thanks to that we were able to quickly establish the persona profile, draft interview questions and use technical terminology with interviewees.