The context
The ride was great.
The operations were expensive.
Coup was genuinely ahead of its time — fully electric, free-floating mopeds in three major cities before anyone else was doing it at scale. Riders loved it. The problem was everything happening after they parked up.
Helmets disappeared. Scooters sat damaged with no one reporting them. Dispatchers coordinated across whiteboards and WhatsApp. Every minute a scooter was offline was revenue lost — and nobody had built the systems to close the gaps.
I was on the ride-phase team: responsible for the experience from finding a scooter to ending the ride. But the real design opportunity wasn't in the app screens riders saw. It was in the invisible layer of operations underneath — and in quietly turning every frustrated customer into a sensor for the fleet.
Problem 01 — Helmet theft
18 helmets stolen daily.
Each one grounded a scooter
for hours.
When a helmet was stolen, the scooter was out of commission until ops sent a replacement. The trouble: operations only found out when a customer complained to support. The lag could stretch to a full working day. Multiply that by 18 stolen helmets and you have a serious, silent revenue leak.
The fix was counterintuitive: don't add a dedicated theft-reporting flow. Instead, repurpose the moment a customer abandons a scooter without starting the engine — a dismissible popup, three taps, gone. It collected the report without feeling like a form. The ops team got an instant flag. The rider barely noticed it.
Before: broken feedback loop keeping scooters offline
After: passive reporting built into the cancellation moment
Problem 02 — 12% cancellation rate
One in eight rides was
cancelled. Nobody knew why.
Damage? Low battery? Wrong scooter? Missing helmet? All plausible — but without data, the operations team was guessing. 12% of rides cancelled before engine start was a significant number. You can't reduce a problem you can't measure.
The same cancellation popup became a data collection engine. By offering structured choices rather than a free-text field, we turned vague frustration into actionable ops intelligence. Within weeks, the team had a prioritised list of real cancellation causes. The guesswork stopped. Fixes could be targeted.
The cancellation popup: one design, two problems solved
"Good ops design is invisible to the rider but worth millions to the business. The best thing I built at Coup was something users barely remembered seeing."
James Ciclitira — on CoupResearch project — community battery swapping
Validating a costly idea
before building it.
The idea: let engaged users earn credits by swapping depleted batteries — reducing ops costs and building community at the same time. The problem: validating it required expensive charging infrastructure that didn't exist yet.
I designed a lightweight app pilot with the internal ops team — small changes that let us collect first-hand data from real users before committing to infrastructure. This is what using design as a research instrument looks like. The answers came back quickly, cheaply, and honestly.
Community battery swapping pilot — research before infrastructure
Operations tooling — the other half of the job
Two tools. One city's
worth of moving parts.
Running a shared fleet means managing hundreds of workers, thousands of vehicles, and constant field exceptions — in real time, across three cities. I was fully responsible for designing and iterating both operational tools.
Web app for ops managers: create and monitor field assignments, track per-scooter status and maintenance history, coordinate workers city-wide in real time.
Mobile app for on-the-ground runners: receive assignments, log maintenance tasks, report scooter status. Designed for one-handed use in the rain.
Enhanced fleet efficiency across all three markets. Fewer hours lost to manual coordination. A documented reduction in company costs from improved uptime.
