Ghanesh RS
ThoughtSpot · Data Portal

Data Portal revamp: a workspace that matched how analysts think

ThoughtSpot was opening up to individual analysts. I led the redesign of the Data Portal, where they connect and model data, from research to a beta launch in three months.

Data Portal revamp: a workspace that matched how analysts think
RoleLead designer
Team2 designers, 2 PMs, 2 engineers, 3 US partners (research, content, design system)
TimelineJul to Sep 2021
PlatformWeb
ToolsFigma, Miro, Loom
Request a live walkthroughView the slides
15+user interviews
5problem areas fixed
3 monthsresearch to beta
FY22Excellence in Design award

In 30 seconds

  • Problem: actions were hidden in one menu, and the portal couldn't grow with new features.
  • Move: a landing page, a persistent left pane, visible creation and one consistent detail layout.
  • Result: a scalable workspace launched in beta, with patterns reused across the product.
Context

ThoughtSpot was opening up to single users.

In 2021 ThoughtSpot launched a self-serve product for individual analysts. The Data Portal, where analysts connect sources, model data and manage worksheets, was built for admins in a sales-led world. If a single user couldn't find their way, they wouldn't stay.

Timeline: from ThoughtSpot Lite in May to a beta of the new data workspace in October 2021.
Timeline: from ThoughtSpot Lite in May to a beta of the new data workspace in October 2021.
The problem

Everything was hidden behind one menu.

  • Low discoverability. Most actions lived under a single overflow menu.
  • Not built to grow. New object types and features had nowhere to go.
  • Out of sync. Admin and Developer pages had moved to a new visual language. The portal hadn't.
The old data portal, with the actions everyone struggled to find marked in red.
The old data portal, with the actions everyone struggled to find marked in red.
Goals

Grow from one user to a whole company.

Business: easy for one user to adopt, then grow into a company license. Product: inviting for analysts and ready for new features. UX: find things fast, simplify, stay consistent.
Business: easy for one user to adopt, then grow into a company license. Product: inviting for analysts and ready for new features. UX: find things fast, simplify, stay consistent.
Research

Interviews and a full audit.

I interviewed 15+ users: internal teams, customer success, and external users of other BI tools. Then I mapped every screen, object and action in the existing portal.

Research takeaways: hidden features, scattered information, too many clicks, confusing names.
Research takeaways: hidden features, scattered information, too many clicks, confusing names.
Audit of the existing architecture: every page, object and action.
Audit of the existing architecture: every page, object and action.
Mapping the mental model: objects, processes and utilities, so the new structure matched how analysts think.
Mapping the mental model: objects, processes and utilities, so the new structure matched how analysts think.
What was broken

Five areas to fix.

Hierarchy: messy structure, many header levels.
Hierarchy: messy structure, many header levels.
Hidden features behind the ellipsis menu.
Hidden features behind the ellipsis menu.
Complex navigation across nested screens.
Complex navigation across nested screens.
Confusing names and redundant terms.
Confusing names and redundant terms.
Constraints

Designing with moving targets.

Requirements

  • Group common functions
  • Quick access to repeated actions
  • Consistent navigation and CRUD
  • Scale for new features
  • Add friction to risky actions

Challenges

  • Untouched legacy features
  • No subject expert for the data app
  • Little usage data
  • Frequent scope changes
  • Tight timelines
Explorations

Home, navigation and page controls.

Home page explorations.
Home page explorations.
Left pane and call-to-action explorations.
Left pane and call-to-action explorations.
Page and filter control explorations.
Page and filter control explorations.
Key decisions

Before and after.

A real landing page

One home for connections, worksheets, tables and views, with the most-used actions up front.

Why: Analysts start with recent work, not an empty list.

Clear navigation

A persistent left pane replaced nested tabs and hidden menus.

Why: Every object type is one click away.

Discoverable creation

A visible Create button with every object type.

Why: No more hunting through an ellipsis menu.

Consistent layouts

Detail pages share one structure across objects.

Why: Learn one page, know them all. New object types slot in.

Grouped features

Related utilities grouped into cards.

Why: Fewer places to look, clearer names.
The design

A mission control for the data engineer.

Landing page.
Landing page.
Object details.
Object details.
New features: SpotApps and integrations.
New features: SpotApps and integrations.
Usage statistics.
Usage statistics.
Utilities.
Utilities.
Testing

Feedback that changed the design.

I tested the explorations with about 10 users before picking one. Customer demos then showed what demo data hid.

Card view for scale

Production accounts had hundreds of connections, so I added a list and card view switcher.

Why: Demo data hid a real scale problem.

Less scrolling

The utilities page was too long. I regrouped it into a compact grid.

Why: People couldn't tell all the features were there.
Outcome

What changed.

Business

Built for organic growth: easier sign-ups and less churn for single users.

Product

Faster time to value through SpotApps and dbt integration. Beta launched October 2021.

Design

Consistent with other product pages, and patterns other teams reused.

Part of the ThoughtSpot work recognized with the Excellence in Design award (FY22).

What I grew in: craft, collaboration and presentation.
What I grew in: craft, collaboration and presentation.