
I was not the original assignee. The project had been briefed to a colleague who reached a point where he could not deliver it alone, and I was brought in at that stage. Arriving without ownership of the decisions already made, I began questioning parts of the brief and the architecture chosen.
Retail store staff across the UK and Ireland who need to serve customers quickly, with minimal training.
That is how the blocker surfaced: retail staff had no corporate Microsoft licences, which made the agreed Power Apps architecture unviable. I called the pivot, re-scoped with the client, and rebuilt on Bubble.io as a mobile-first no-code build.
The design phase led with a mobile-first layout in Figma, the interface had to work on a smartphone or in a narrow desktop window, running alongside other tools on a busy shop floor.
In development, I used Bubble's low-code environment to build the authentication and workflow layer — two-factor authentication, password resets, and a custom basket management system for tracking spend against voucher limits. I also built a separate admin page so project managers could approve or remove users without touching the underlying database. I ran usability testing across 15 core tasks — sign-up, OTP requests, editing the basket — to make sure the flow held up under real retail conditions.
In development, I used Bubble's low-code environment to build the authentication and workflow layer — two-factor authentication, password resets, and a custom basket management system for tracking spend against voucher limits. I also built a separate custom admin page for Canon stakeholder could approve access requests directly. I ran usability testing across 15 core tasks — sign-up, OTP requests, editing the basket — to make sure the flow held up under real retail conditions.
I ran it to departure. Voucher lists arrived monthly from the third party Canon used to generate them and were applied manually, I intended to automate this, but that vendor had no shareable API, so the constraint was external rather than technical. Alongside that: access administration, periodic issue checks, and usage reporting. Reporting is worth a note, we proposed it for v1, the client declined, I built it afterwards, and once live it was adopted and became useful to them.
In production across UK and Irish retail teams for over a year, still running at departure. No adoption data exists: Canon ran the retail training and rollout internally and kept the numbers.