the problem
Five channels each ran their own upsells, with no shared vocabulary, measurement or strategy.
popeyes · 2026 · ux design
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.
the two-minute version
what i did
I audited every surface, benchmarked 9 brands and built one framework: when to ask, what to offer, how to rank.
the result
+18% upsell purchases in a live A/B test, and a taxonomy leadership adopted company-wide.
- upsell purchases in a live A/B test
- +18%
- qsr brands benchmarked
- 9
- participants said hidden prices hurt trust
- 38 of 39
the problem
every channel spoke a different upsell language
Each channel ran its own upsells, with no shared vocabulary, no shared measurement and no strategy. Many app users never scrolled far enough to see the first upsell.
Every channel spoke a different upsell language: different treatment, framing, timing and form.
the process
audit everything before designing anything
I audited every upsell surface in 5 channels and benchmarked 9 QSR brands across 6 funnel stages. What I found: everyone has solved the front of the funnel. The difference is in the cart, checkout and after the order.
Auditing every upsell across four channels made the gaps obvious before we designed anything. three questions: when, what, and in what order
I built the framework around three questions: when to ask, what to offer, and how to rank.
Rank, not offer, decides which upsell wins, so slot order became a design decision.
the solution
one model, every channel keeps its own screen
A 4-type upsell taxonomy (Size Up, Upgrade, Premium Swap, Add-on) with triggers, and a map of 15 surfaces.
One model ranks the offer; each channel keeps its own UI. badges that only say true things
An upsell card component with 84 variants, and social-proof badges grounded in Cialdini. The rules: one badge per card, and every claim has to be true.
The same badge language on the app and the kiosk.
impact & next steps
+18% upsell purchases, and a shared language
In a 2026 live A/B test, my recommended checkout pattern lifted upsell purchases 18%, a statistically significant result, and became the production default.
Leadership adopted the taxonomy company-wide, across app, kiosk, POS and marketing, and added 3 of my team’s projects to the next roadmap.
Four independent levers, each able to win alone and combine with the others.
the call i made: keep the prices
Hiding prices looked like an easy way to lower hesitation. Research predicted it would hurt trust instead: 38 of 39 participants said so. The live test confirmed it, and prices stayed.
more from the project
Scroll sideways for the screens that didn’t fit the story.
next up
checkout research: 96 sessions
Two rounds of factorial usability testing that shaped a better checkout, plus a code-first research setup that stopped broken prototypes from losing data.
or see all 20 projects