Back to work
Periscope Data

Interactive Data Analysis

Users wanted to analyze data without knowing SQL, so we provided support for interactive analysis that became a highly adopted feature.

Timeline
December 2017 – May 2018
Team
UX Designer, PM, Engineering Lead
Feature
Data Discovery
Building a daily revenue area chart in the Discovery editor
Building a chart by choosing fields and an aggregation.

Background

Periscope Data helps data teams answer critical questions, increasingly by opening data up to everyone. This project was a chance to empower salespeople, marketers and other business users, and to save analysts time by operationalizing their work.

The problem

Analysts hit an upper limit

Analysts could only build so many dashboards and ad-hoc analyses. Self-service for partner teams would free them for more advanced work.

Business users left the product

Without SQL, many downloaded CSVs into Excel or turned to competing tools to answer simple questions.

Research

Before kicking off, I ran remote user interviews to understand business users’ needs, how they work with data teams and IT, and how comfortable they are joining data. These findings shaped the PM’s 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.

Many users already thought in drag and drop, so we focused on testing that interaction. I also researched BI tools to understand industry standards.

Wireframes, testing and visual design

After rounds of user, stakeholder and engineering reviews, I wireframed for usability testing, exploring:

  • How users think about creating an analysis
  • Auditing an interactive chart
  • Charts of mixed origin
  • Aggregation, filtering and limits

Once the direction was approved, I designed the visuals, building on our existing visual language.

Three Discovery editor windows: table builder, SQL editor and calculated field dialog
Editor states, from the table builder to calculated fields.

Interactions

The central interaction was code-free analysis. I explored dragging into designated fields, typing into fields, and dropping a field onto a chart’s X and Y axes. Users preferred the last by far, but it was the hardest to build, so we scheduled it for a later release.

Animation of dragging fields into the chart editor
Dragging fields into the chart editor.
Typing to find a field in the Y axis
Typing to find a field.
Dragging a column to reorder it in a table
Dragging to reorder columns.
Dataset preview card showing a description, owner and tags
Previewing a dataset before choosing it.

Internal testing and release

We released internally and gathered feedback in a Slack channel, plus a contest for the best dashboard. I worked with engineers to catch edge cases, and after launching as Data Discovery, the PM and I joined customer feedback calls.

Chart editor with a toggle between SQL and Discovery
A toggle above the schema tab switches between the SQL and Discovery editors.

Constraints

Business users’ experience had to remain accessible to analysts, so layout decisions followed the existing editor. Users preferred starting from the familiar new-chart flow on a dashboard, and since analysts needed both editors, we added a toggle above the schema tab.

Results

40%+

Most high-value accounts had Data Discovery enabled, and over 40% of those adopted the new workflows within six months. Sales reps also credited it with helping close renewals and new business.

Lessons and challenges

Discoverability

Prioritizing analysts’ path to query building meant compromising feature visibility. We planned to address this through information architecture work.

Work with an area expert

The PM’s analytics background helped us quickly define the persona, draft interview questions and use the right terminology.