01

Summary

Prism's analytics dashboard was powerful — but so complex that 63% of users never made it past the first screen.

Product Design UX Research Prototyping Design System Usability Testing

We were brought in to redesign the core dashboard experience for Prism Analytics — a B2B SaaS platform serving 50,000+ data analysts. The product had accumulated years of feature debt, and users were bouncing without ever getting value.

Over 14 weeks, we ran a full research-to-production process: interviews, heuristic audits, competitive analysis, multiple prototype rounds, and two usability test cycles. The result was a completely new information architecture and interaction model that reduced time-to-insight by 54%.

67%
Task Completion ↑
54%
Time to Insight ↓
4.6
SUS Score (was 2.9)
28%
Churn Reduction
02

The Problem

Prism had been built by engineers for engineers. Every feature request over four years had been bolted on without a coherent design strategy. The result was a product that could technically do everything — but practically helped no one.

Support tickets were dominated by "where is X" and "how do I do Y." Power users had created internal wikis just to explain the interface to new teammates. Churn interviews consistently pointed to the same culprit: the dashboard.

01
Information Overload
47 visible controls on the default view. Users reported feeling overwhelmed within 30 seconds of logging in.
02
No Clear Entry Point
New users had no guided path. The blank state was a blank page with a blinking cursor and zero scaffolding.
03
Inconsistent Patterns
Similar actions had 4 different interaction patterns across the app. Users could never build reliable mental models.
03

Users & Research

We ran a 3-week research sprint before touching a single wireframe. Eighteen interviews, one diary study, and a heuristic audit across every major flow.

A
Anya Sharma
Senior Data Analyst, 6 years exp.
"I know it can do what I need. I just can't find where to start every single time."
M
Marcus Chen
BI Manager, leads team of 8
"Onboarding a new analyst used to take two weeks. The product just isn't self-explanatory."
P
Priya Nair
Startup Founder, non-technical
"I paid for this to get answers. I'm just getting more questions."
18
User Interviews
60 min each
↳
Diary Study
7 days, 12 participants
∿
Heuristic Audit
Nielsen's 10 heuristics
◈
Analytics Review
6 months session data
04

Frustrations &
Findings

Research surfaced six recurring patterns — each one a compounding failure in the user experience.

01
Cognitive Load at Entry
100% of new users reported feeling confused within 60 seconds. The dashboard loaded in a state optimised for no one.
Critical
02
Broken Filter Logic
Filters didn't persist across views. Users re-applied the same filters an average of 7 times per session.
Critical
03
No Contextual Help
Zero in-product guidance. Users either figured it out alone, watched YouTube tutorials, or abandoned the task.
High
04
Invisible Data Relationships
Charts didn't communicate how data points related. Analysts had to export to Excel just to understand correlations.
High
05
Collaboration is an Afterthought
Sharing a view required copying a URL, pasting it in Slack, and hoping the recipient had the same filter state.
Medium
06
Mobile Unusable
22% of users tried to access reports on mobile. None of them succeeded. The layout was a direct desktop port.
Medium
05

Redesign

The redesign wasn't a facelift — it was a complete rethinking of how users navigate, explore, and act on data. We started from the user's job-to-be-done, not the existing feature set.

New dashboard overview
Before → After
Consolidated dashboard — from 47 controls to 9 primary actions
Chart explorer
Chart Explorer
Progressive disclosure for complex data
Filter panel
Persistent Filters
Context preserved across all views
01
Progressive Disclosure
Show the 20% of features that cover 80% of use cases. Reveal complexity only when the user signals readiness.
02
Persistent Context
Filters, time ranges, and view preferences persist everywhere. Your last state is your next starting point.
03
Guided Exploration
Smart defaults and contextual suggestions reduce the blank-slate problem for both new and returning users.
04
One-click Sharing
Share any view — with full filter state — in a single click. Recipients land in the exact same context.
06

User Testing

Two rounds of moderated usability testing — 8 participants each — on Figma prototypes at 60% and 95% fidelity. We measured task completion, time-on-task, and SUS scores.

67%
↑ from 31%
Task Completion
54%
↓ time saved
Time to Insight
4.6
↑ from 2.9 / 5
SUS Score
0
↓ from 14 avg.
Rage Clicks / session

"I just found what I needed in under a minute. I've never been able to do that before."

AS
Anya S. — Senior Data Analyst, Round 2 participant

"The sharing alone would save my team hours every week. This changes how we collaborate."

MC
Marcus C. — BI Manager, Round 2 participant
07

Snippets

Research
Research
Stakeholder interviews & kickoff
Affinity map
Research
Affinity mapping & insight clustering
Wireframes
Wireframes
Low-fidelity layout explorations
IA map
IA
Information architecture map
Dashboard
Design
Dashboard redesign — final HiFi
Filter panel
Design
Filter panel & persistent state
Component library
Design System
Component library — 140 atoms
Usability testing
Testing
Remote usability session — Round 1
Handoff
Handoff
Dev handoff documentation
08

Impact &
Learnings

The redesign shipped to 100% of users 6 weeks after handoff. Results were measurable within the first billing cycle.

67% increase in task completion rate
Measured via session recording across 2,400 users in the first 30 days post-launch.
28% reduction in monthly churn
Churn interviews now rarely mention the dashboard as a pain point.
Support tickets down 41%
"How do I…" tickets dropped sharply. Contextual help eliminated the most common queries.

Key Learnings

01
Ship research artefacts, not just outputs
Sharing affinity maps and raw quotes with the client team created alignment that a slide deck never could. They started advocating for users internally.
02
Test earlier than feels comfortable
Our round 1 prototype was "too rough" to show stakeholders. Participants didn't care — they surfaced critical flaws that would have cost weeks in round 2.
03
Defaults are design decisions
The blank state, the default time range, the first chart shown — these are as important as any feature. We spent a week just on defaults. Worth it.
09

What's Next

Phase 1 is live. Here's what the roadmap looks like for the next two quarters, based on post-launch feedback and product analytics.

01
AI-powered insight summaries
Automatically surface anomalies and trends, written in plain language. Designed but awaiting ML infrastructure.
In Progress
02
Mobile-first reporting view
A dedicated mobile experience for executives who need quick read-outs on the go. Currently in wireframe phase.
In Progress
03
Collaborative annotations
Leave notes and highlights directly on charts, taggable by team member. Scoped and designed, queued for Q3.
Planned Q3
04
Custom dashboard builder
Drag-and-drop dashboard creation for power users who want to compose their own views from existing chart types.
Planned Q4
Have a similar
challenge? Let's talk.
Start a Project See more work