the problem
Crews had no real-time feedback on upselling, and two data sources disagreed on what even counted as one.
popeyes · 2026 · design + code
poppy pulse
A performance tool on the restaurant register that turns upselling into a team game. I started it myself and took it from concept to field research to production code.
the two-minute version
what i did
I defined one metric, built a prototype that feels like the real register, tested it in 4 restaurants and wrote the production front end.
the result
Shipped as production code, demoed at the franchisee convention, and handed to engineering for a pilot.
- restaurants in the field study
- 4
- sessions with crew and managers
- 6
- prompts per order, at most
- 2
the problem
crews were upselling blind
Most of a quick-service restaurant’s sales go through the drive-thru and the front counter, but crews had no real-time feedback on how they were doing. Upsells weren’t measured at the register, and leadership saw register prompts as too complex to take on.
Crews already tracked upsells by hand on the wall, so we brought that scoreboard onto the POS.
the process
one number everyone agrees on
Two data sources disagreed on what counted as an upsell. I defined a single metric that reconciled them. Every number on screen got a definition, a formula and a source.
Every number on screen derives from four inputs, so the screens can't contradict each other. a prototype that feels like the register
I took hands-on POS training, then built a working browser prototype that looks and behaves like the register crews use every day.
I set up a study in two days: 6 sessions across 4 restaurants. I played the customer while crew members and managers used the prototype.
The prototype flow we tested with crew in four restaurants.
the solution
the bar is the product
A progress bar that stays on the POS all shift. The Fogg Behavior Model, loss aversion and small celebrations shaped every piece of it.
The bar is the product: shift progress lives on the POS, one tap from the full view. two prompts at most, timed to the order
At most 2 upsell prompts per order, timed to what’s in it, and a small celebration each time an upsell lands.
Every prompt explains why it fired and gives crew one natural line to say. a leaderboard of stores, not people
A manager dashboard with a store leaderboard. Rankings are by store only, to protect crew privacy.
One shift, one upsell rate, and a next target crew can act on.
impact & next steps
shipped in production code I wrote
I shipped the UI as production code: a merged pull request to the POS extension, and a dashboard pull request that passed every automated and visual check. I caught an access-control gap before release.
At the franchisee convention, franchisees and general managers “were super excited about it.” It’s now with engineering for a pilot.
the call i made: rank stores, not people
The leaderboard ranks stores, never individual crew members. Upselling becomes a team goal, and crew privacy stays protected.
more from the project
Scroll sideways for the screens that didn’t fit the story.
what i learned
Building variants beats debating them. Field research with crew changed the design more than any review did. And designers who ship code get to decide how the details turn out.
next up
cross-channel upsell strategy
One shared language for upselling across the app, kiosk, voice AI, POS and menu boards, tested through research and live experiments.
or see all 20 projects