Back to Portfolio
SaaS Dashboard UX Data Visualisation Shipped · 2024

SaaS Analytics Dashboard:
From Overwhelming to Effortless

Redesigning a complex, 140-metric analytics dashboard for 50,000+ B2B users — reducing cognitive load by 55% while increasing daily active engagement by 38%.

My Role
Senior UX Designer
Duration
8 Months
Team
1 UX, 1 PM, 6 Eng
Platform
Web SaaS (B2B)
Tools
Figma · Hotjar · Maze
55%
Cognitive Load ↓
38%
Daily Engagement ↑
50K+
B2B Users
4.7★
User Satisfaction

The Product

A B2B SaaS analytics platform serving marketing, operations, and finance teams at mid-market and enterprise companies. The dashboard was the product's primary surface — the first screen users saw every day and the place they spent the majority of their time.

Over 4 years of growth, the dashboard had accumulated 140+ metrics, 12 chart types, and 8 filter panels — each added reactively to satisfy individual customer requests. The result was a screen that tried to do everything and succeeded at nothing.

"I open the dashboard every morning and I honestly don't know where to look. There's so much data I just end up ignoring most of it."
— Operations Manager, Enterprise Client (User Interview)

140 Metrics, Zero Clarity

Quantitative data confirmed what users were telling us anecdotally. Hotjar session recordings showed users spending an average of 47 seconds on the dashboard before navigating away — a strong indicator that users weren't finding the information they needed quickly enough to justify staying.

✗ Before — The Chaos
140+ metrics displayed simultaneously with no visual hierarchy
8 separate filter panels, each with different interaction patterns
No concept of "important vs. contextual" data — everything treated equally
Average time to find a specific metric: 2 minutes 18 seconds
Mobile completely unusable — horizontal scroll required to reach key data
No personalisation — same dashboard for a junior analyst and a CFO
✓ After — The Solution
Progressive disclosure: 6 key metrics visible by default, 140 accessible on demand
Single unified filter bar with consistent interaction pattern
AI-assisted "smart highlights" surface anomalies and trending metrics
Average time to find a metric: 18 seconds (87% improvement)
Fully responsive — key metrics accessible on mobile in one scroll
Role-based views + user-configurable widget system

Understanding Decision-Making Patterns

The core research question was deceptively simple: What do users actually need to see first thing every day, and what can wait? The answer, it turned out, varied dramatically by role — which led to our progressive disclosure and role-based personalisation strategy.

🎥
Session Recordings
200+ Hotjar session recordings analysed. Found that 78% of users navigated away from the dashboard within 60 seconds — and that 3 specific metrics accounted for 65% of all scroll and click interactions.
👤
Role-Based Interviews
24 interviews across 4 user roles (analysts, managers, executives, ops teams). Each role had completely different "first questions" they needed the dashboard to answer immediately.
🗺️
Task Analysis
Mapped 40 common dashboard tasks across all user roles, then scored each for frequency and importance. This matrix became the direct input for prioritising visible vs. secondary metrics.
📉
Funnel Analytics
Product analytics revealed a 34% weekly churn from the dashboard to a third-party tool. Users were exporting data to Excel because the dashboard couldn't give them the views they needed fast enough.

The most important research finding was the "3 questions" pattern: regardless of role, every user opened the dashboard to answer one of three questions — "How are we doing today?", "What changed since yesterday?", and "What needs my attention right now?" The redesign was built around these three core intents.

Progressive Disclosure as the Core Model

The defining strategic decision was to adopt progressive disclosure as the fundamental UX pattern. Instead of showing everything at once, we designed a three-layer information hierarchy:

Layer 1 — Headline view: 6 role-appropriate KPIs visible immediately on load, with trend indicators and smart anomaly highlights. No scrolling required on any device.

Layer 2 — Standard view: Full dashboard with 24 configurable widgets, expanded charts, and the unified filter system. One click from the headline view.

Layer 3 — Deep dive: All 140 metrics accessible through a searchable metric library, with the ability to build and save custom views. Designed for power users and analysts.

01
🎯
Intent-First
Every design decision anchored to the 3 core user intents identified in research
02
📊
Progressive Disclosure
Show what matters most first; reveal depth progressively on demand
03
🎭
Role Adaptation
Defaults personalised by role; full customisation available to all users
04
📱
Mobile-First
Headline view designed for mobile first, then scaled up for desktop

A Dashboard That Thinks With You

The redesigned dashboard opened on a clean, breathable headline view. The visual hierarchy was completely rethought: large, scannable KPI numbers with clear trend indicators at the top; a primary chart relevant to each role centre-stage; smart highlights calling out anomalies the user should know about.

Analytics Overview · Today
Last 7 days ▾
Export
Revenue
$84.2K
↑ 12.4% vs last week
Active Users
12,481
↑ 8.1% vs last week
Churn Rate
2.1%
↓ 0.4pp vs last week
NPS Score
67
→ unchanged
Revenue Trend — 7 Days
⚡ Smart Highlight
Revenue up 12.4% — driven by Enterprise tier conversions this week.
Fig. 1 — Redesigned headline dashboard view. 4 role-appropriate KPIs, a primary trend chart, and a smart highlight panel — all visible without scrolling on any device.

The Smart Highlights panel was a particularly impactful addition. Powered by simple statistical anomaly detection (not ML — deliberately simple to ship quickly), it surfaced the most significant changes since the user's last session. User testing showed this single component reduced the average time-to-insight from 2m18s to 18 seconds.

Validating Every Layer

We ran three rounds of usability testing across the 8-month project — one per major milestone. Testing was conducted via Maze for unmoderated sessions and Zoom for moderated sessions, with a mix of existing customers and recruited participants matching our target personas.

📋
Round 1 — IA Validation
Card sorting with 18 participants to validate the progressive disclosure hierarchy. Found that "smart highlights" resonated strongly but the initial label "Insights" caused confusion — renamed to "What Changed".
🎨
Round 2 — Prototype Testing
Hi-fi prototype tested with 12 existing users via Zoom. Identified that the "Layer 2 → Layer 3" transition was unclear — users couldn't find the metric library. Redesigned the entry point based on this finding.
🚀
Round 3 — Pre-Launch Beta
250 beta users tested the near-final build over 2 weeks. Net Promoter Score for the dashboard jumped from 23 to 61 compared to the old version — before full public launch.
Accessibility Audit
Full WCAG 2.1 AA audit conducted with an external specialist. 14 issues identified and resolved before launch — including colour contrast failures in the chart legend system and missing ARIA labels on interactive chart elements.

What Changed for 50,000 Users

The redesigned dashboard launched in phases over 6 weeks. Post-launch metrics were measured at 30, 60, and 90 days against the pre-launch baseline.

87%
Faster time-to-insight (2m18s → 18s)
34%
Drop in Excel exports (users staying in-app)
NPS 61
Dashboard NPS (was 23)

Beyond the metrics, the redesign created a platform for the product to grow. The widget system and role-based views established a flexible architecture that the engineering team has since used to ship 3 new feature areas without requiring UX rework of the dashboard foundation — a significant multiplier on the original design investment.

"For the first time in years, I actually look forward to opening the dashboard in the morning. I know what I need to know within 30 seconds. That's completely new."
— VP of Marketing, Enterprise Customer (Post-launch interview)
Related Case Study
Oracle ERP → Cloud Migration UX Overhaul