let's talk
popeyes · 2026 · ux research

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.

the two-minute version

the problem

We needed to know which mix of price display, bag design and badges helped guests at checkout.

what i did

Two factorial studies, 96 sessions, and one coded prototype with 8 variants in place of six Figma files.

the result

The most common complaint went to zero in 55 sessions, and clear rules went into the checkout.

my role
designed and ran both studies, built the coded prototype
platform
App checkout
deliverables
Study design, Screener, Coded prototype, Findings, Checkout rules
sessions in two rounds
96
variants in one coded prototype
8
top complaints in 55 sessions
0

the problem

  1. prices alone didn’t move it

    We needed to know which mix of price display, bag design and badges helped guests at checkout, and which got in their way.

    Slide comparing a control and two price-visibility variants of a combo page.
    A 2025 test showed that adding prices alone didn't help, so we looked deeper at checkout.

the process

  1. round 1: people lost track of the total

    A 2 × 3 design, 40 unmoderated sessions. A stepped checkout buried the cart total: half the people in that version couldn’t find it, and nobody had that problem in the others. A sticky running total fixed it.

    Scorecard comparing three checkout variants.
    Length wasn't the defect; losing track of the total was.

  2. round 2: one coded prototype, eight variants

    A 2 × 2 × 2 design (price, sticky bag, badges), 56 sessions, with screener quotas based on real order demographics.

    Research finding slide with a sticky bag total treatment.
    Two rounds, same defect: make the total visibly change on add.

the solution

  1. rules the checkout now follows

    Upsell items filtered by what’s already in the cart.

    At most 3 carousels, sauces first, never re-show what’s already in the bag, always show price and calories.

    Three-column recommendation grid.
    What the research told us to ship, show and fix.

impact & next steps

  1. the top complaint went to zero

    Filtering by what’s in the cart took the most common complaint from frequent to zero in 55 sessions. Every state, flow and new component went to engineering on one page.

    Large board of final app screens and components.
    The engineering handoff: every state, flow and new component on one page.

the call i made: one coded prototype, not six figma files

Broken prototype links had cost us data in round 1. For round 2, I replaced six hand-built Figma prototypes with one coded prototype that switches between 8 variants from the URL.

more from the project

Scroll sideways for the screens that didn’t fit the story.

Four phone screens showing the full app order flow.
The default combo path, end to end.
Side-by-side recommendation carousels on app and kiosk.
Multi-pick: guests can add more than one recommendation without leaving the flow.
Badge set with display logic and control vs. test carousels.
A defined badge library with strict rules: one signal per card, at most two per carousel.
Stepped checkout on app and kiosk with a logic table.
The checkout bag becomes a short guided sequence, one decision at a time.
Control and test recommendation cards on app and kiosk.
Testing whether removing labels at the moment of add lowers hesitation.

what i learned

Compare what people say with what they do. Say so when the test setup fails. And read results against total sales, not just upsells.

next up

kiosk menu redesign

A kiosk menu guests can find their way through, tested with a three-version coded prototype and two rounds of multi-cohort testing.

or see all 20 projects
ux research · popeyes

kiosk menu redesign

ready to make something awesome together?

let's work togethersee my resume
say hello download resume