Project detail
Navigation Redesign for B2B SaaS
Rebuilt the navigation and information architecture of Hala Mobility’s B2B fleet platform, cutting the time to reach key modules from ~9s to ~5s.
Role: Led the navigation and IA redesign, as one of a team of four designers
Product: Hala Mobility · CercleOS, a B2B fleet management platform (web)
Users: Sales, operations and customer care teams
Timeline: 2024 – 2025
Methods: 50+ screen audit, heuristic evaluation, interviews, competitor benchmarking, card sorting, usability and A/B testing
TL;DR
Problem: CercleOS had grown an icon-only sidebar of nested, overlapping menus. Teams couldn't find daily tools, and new hires spent more time learning navigation than the job.
What I did: Rebuilt the sidebar, tabs, breadcrumbs and IA around how teams actually group their work, validated with card sorting, usability tests and an A/B test.
Impact: Time to reach a key module ~9s → ~5s · drop-off after 2+ clicks 14% → 6% · misclicks down 61%.
Built by developers, outgrown by its users
CercleOS is Hala Mobility's B2B platform for managing the EV lifecycle: vehicles, batteries, rentals, riders and payouts. It was built by developers with little focus on UX or IA. As the product scaled, every new feature became another icon in the sidebar.
I took the initiative to lead the overhaul: I audited 50+ screens, ran a heuristic evaluation of critical workflows, and redesigned the sidebar, breadcrumbs, in-page navigation and key workflows with stakeholders, developers and internal teams.
Where navigation broke down
Confusing labels: modules like In-Transit and Recovery Vehicles were hard to tell apart.
Slow workflows: nested menus and extra clicks slowed daily operations.
No room to grow: new features didn't fit the existing menu.
Hidden actions: important workflows were buried.
Baseline from web analytics: ~9s on average to reach a key module, 14% drop-off after 2+ clicks, and frequent misclicks.
What users told us
I interviewed the customer care team, 6 hub managers and 3 internal ops team members, asking how their day starts, what they open most, and what they never touch.
"When I'm training new people, I spend more time teaching them how to navigate than teaching them the actual job." – Senior Operations Lead
High-frequency tasks were hard to find.
The sidebar was overwhelming, so people used only a fraction of the features.
Navigation slowed onboarding for new hires.
Borrowing what already works
I benchmarked Coulomb AI, Fleetio, Samsara and Fleetx, and kept only the patterns that suited CercleOS's depth.
Decision 1: Group by how teams work, not by how the database is built
Options: keep grouping by technical module, which mirrors the backend and is easy to maintain, or group by the workflows people actually run.
I chose workflows, and let the teams decide them. Stakeholders from Operations, Tech and Management sorted module cards (Vehicles, Drivers, Trips, Battery, Charging, Recovery, Reports, Billing, Settings and more) into themes and named them. That gave us clear groups, labels in the teams' own language, fewer duplicate terms, and buy-in before a single screen was designed.
Decision 2: Labels on every item, in a sidebar that gets out of the way
Options: keep the compact icon-only rail, or show labels and group headings.
I chose labels plus icons in expandable groups, because nobody should have to guess what an icon means. The cost: a wider sidebar takes space from dense data tables. So the new sidebar responds to cursor movement and collapses when you're working, giving the data room back.
Decision 3: Three navigation levels, one job each
Mapping critical flows (assigning vehicles, tracking recovery, checking battery lifecycle) showed repeated steps and inconsistent navigation between levels. I combined related actions and set up three levels: the sidebar for where you are, tabs for which slice of a list you need, and a detail view for one record, with breadcrumbs leading back up.
The new CercleOS
Results
Web analytics (live): time to reach a key module ~9s → ~5s; drop-off after 2+ clicks 14% → 6%
Usability testing (8 users, task-based): reaching a module within 5 seconds 64% → 92%; average time to find a page 12s → 6s
A/B test: misclicks down 61%; daily use of deeper modules like Recovery Vehicles up 22%
What I learned
Let the users sort the menu. Card sorting settled naming debates that would otherwise have gone on for weeks, and it created buy-in for free.
Multiple roles mean multiple mental models. One-size-fits-all navigation breaks quickly in B2B software.
Clarity drives adoption. The +22% use of deeper modules came from features people could finally find, not new features.
Want the full walkthrough? I'm happy to take you through it in an interview.
