Two quarters into a loyalty app redesign, I noticed the build quality was patchy from one release to the next. One release, the home screen came back close to pixel-perfect. The next, the redemption flow looked stitched together in a hurry, inconsistent components, mismatched fonts, abrupt jumps between screens. I wanted to know why the same design squad produced such uneven results.
I asked the product managers to gather feedback from the mobile engineers, a short form: what worked, what went wrong, what could improve. The answer that kept coming back was that the design files were hard to follow. We had been putting every feature into a single Figma file, every flow from the common path to the edge cases, organised for completeness so the whole team could see how a feature worked end to end. To an engineer picking up one sprint’s tickets, that completeness read as noise.
So I sat with the design team and we changed how we delivered. We reorganised the files by sprint, cross-referenced each screen to its PRD section and ticket, and cut each handoff down to only what that sprint needed. For a few weeks the build quality improved. Then, by the end of the quarter, it started to slide back.
The regression was the part that taught me something. I went back to the engineers, and this time the real constraint emerged. They were close to burnout, split across several projects, dividing every day between business logic and interaction logic. When the week got tight, they shipped the business logic and let the interaction detail go. Given their load, that was the rational choice.