Enter password to view case study
Context
N!mbus was losing customers after demos. Sales could sell the product, but users couldn't navigate it independently—leading to significant post-demo drop-off (exact rate tracked by sales team), high churn risk, and growing support costs.
I led a full redesign to close this gap: rebuilt the information architecture, designed structured onboarding, improved accessibility (WCAG 2.1 AA), and established a design system from scratch.
Before:
Users needed 3-4 sales calls to understand basic filtering
After:
New users filtering properties in first session without support
Before
N!mbus had 5,000+ paying users—but 63% had never discovered its most valuable feature
While sales demos worked, users struggled alone. They couldn't find location analysis even after months of use, filters were buried behind popovers with no visibility of what was applied, and the site info panel required multiple clicks just to see basic property data. Without clear navigation or trust signals, users abandoned the platform or defaulted to spreadsheets to validate decisions. In a high-stakes property market, this hesitation meant lost deals and churn.
N!mbus before redesign
Dense left navigation with unclear priorities • Data-heavy panels without visual hierarchy • Map layers buried in nested menus • No clear entry point for new users
Constraints & trade-offs
Why redesigning this was risky
N!mbus was already making money. It had 5,000 paying users. Which meant every design decision could cost us revenue. Change too much, existing users churn. Change too little, new users bounce. And I had to figure this out with barely any research budget and requirements that kept shifting.
The initial redesign took 8-10 months—design system from scratch, filters, location analysis, saved sites, info panel. Then ongoing evolution for 2.5 years: comparables, letter generation, expanded layers. And throughout all of it, I maintained the legacy system in parallel. Every new feature needed to work in both systems until we could fully migrate.
Engineering wanted to ship incrementally. Sales wanted everything at once for demos. I couldn't give both what they wanted, so I built high-fidelity prototypes that sales could demo while engineering worked at their own pace. Sales got clickable designs to show prospects. Engineering kept quality without rushing. The prototypes closed deals—contributing to the 56% ARPU increase—before a single line of code shipped.
Research approach
Combined qualitative research (user interviews, usability sessions with screen recordings) with behavioral data analysis and affinity mapping to identify patterns across diverse user types.
The turning insight
Confidence - not features - was blocking decision-making
Across interviews, usability sessions, and behavioral data, one pattern emerged: users hesitated because they didn't trust what they were seeing. They needed to understand what they were looking at, why it is important, and whether it was safe to act on - without a sales call.
“I didn’t realise map layers (Location analysis) were included - even after a year of using the product.”
Existing user
“I wasn’t sure whether it missing data - or if the system was still loading.”
Existing user
“I needed clearer signals to understand what each action would actually do.”
Existing user
Execution under constraints
Designing clarity in a moving system
With shifting requirements and competing priorities. I couldn't wait for perfect research, so I prototyped fast and tested rough. Quick usability sessions. Clear choices about what to sacrifice.As the redesign progressed, I established a design system to prevent new complexity.
Exploring filter patterns
Organising 40+ filters without overwhelming users required testing multiple approaches. I explored how different patterns affected the speed and confidence of property searches. Tested with 6 users. Version 2 failed because people couldn't pick a role—'I'm both a surveyor and an investor, which tab do I use?' They spent more time deciding than filtering..
😶
So-so
All filters visible
All filters expanded by default. Every option immediately accessible.
❌ Users overwhelmed by 40+ options
❌ Couldn't distinguish basic from advanced
❌ Abandoned before filtering
👎
Failed
Role-based tabs
Filters organised by user role (Developer | Surveyor | Investor).
❌ Users didn't identify with role labels
❌ "I'm both a developer AND surveyor—which tab?"
❌ Created barriers instead of clarity
🏆
Winner
Progressive disclosure
Collapsible categories + keyword search + background processing.
✅ Users started filtering immediately
✅ Search revealed advanced options when needed
✅ Favourites for frequently used filters
From features to decisions
Designing for real decision-making
N!mbus had grown feature by feature. I mapped the real decision journey users followed—scanning, building confidence, validating constraints, taking action.
Users included developers, surveyors, investors, and planners with different goals. I focused on patterns: progressive disclosure, site cards for quick scans and deep dives, mobile views for on-site decisions. This reduced cognitive load without sacrificing depth or speed.
Desktop and mobile views
Comprehensive property data accessible on-site via mobile or during desk research on desktop. Designed for quick scanning and deep analysis depending on user context.
The moment of truth
Would 5,000+ users accept the change?
Redesigning a live product with paying customers meant every change carried risk. We rolled out the new filters first to measure impact.
The result: Support tickets dropped 36% in the first two months. A year later, that reduction held—users kept finding what they needed without asking for help.
The prototype that enabled sales conversations
Sales used this clickable prototype to demo the new filter experience before development. Prospects wanted to buy what they saw—driving the 56% ARPU increase.
After
A product users could navigate and trust on their own
Two major improvements shipped while broader evolution continued. Sales leveraged detailed prototypes to close deals on future capabilities, generating revenue before launch. Capabilities became discoverable and trustworthy, with users relying less on sales walkthroughs
Business impact
From sales dependency to product-led growth
High-fidelity prototypes became a sales tool before launch—prospects could visualize the value and commit to future capabilities, generating revenue before a single line of code shipped.
Within one year (FY2024 → FY2025):
~2× ARR growth
+56% average contract value
33% reduction in user friction, 25% increase in usability satisfaction, 3× growth in advanced feature adoption
36% reduction in support tickets, freeing ~40% of sales capacity for expansion vs. onboarding
"The new interface finally makes sense—I found features I didn't know existed that are saving me hours every week."
— Surveyor, existing customer
"For the first time, I didn't need to call support during my trial."
— Residential Developer,
new customer
"We can finally upsell advanced packages because customers can actually use them."
— Sales Director
What this taught me is.
Every challenge can show you more perspectives
Trust isn't designed—it's architected. You can't polish your way to confidence when the foundation is unclear. The filters worked because they showed everything upfront, then let users collapse what they didn't need. Learn first, optimize later.
Personas are fiction. Real users don't fit in boxes. Give them control instead.
And sometimes your best sales tool isn't a feature—it's the prototype that shows what's possible.
What's next: AI-powered filters. Smart recommendations. Self-serve onboarding. Same principle: meet users where they are, not where the system expects them to be.
