Role: UX Consultant and Product Designer, eight-week engagement, working as the sole designer alongside a product manager and a small mobile engineering team.
The redesigned flow reduced pickup-location complaints by approximately 42% during the six-week pilot, while mobile-order adoption at participating locations increased 16%.
Context and Constraint
Customers were arriving at the wrong pickup point, forcing restaurant staff to locate orders, redirect customers, and sometimes remake food. A four-week review found pickup-location confusion affected roughly 3% of mobile pickup orders, creating delays, complaints, and extra staff intervention. Because the existing POS and kitchen-routing infrastructure could not be substantially changed, the fix had to happen in the mobile ordering flow.
The Reframe
Evidence
Decisions and Tradeoffs
The redesign deliberately added 1-4 seconds. It sounds counter-intuitive, but for a reason: users had to explicitly choose Drive-Thru Window, Dine-in, or In-Store Counter, with no default selected. We also replaced vague labels with language that described the actual handoff point. An early concept reused the customer’s previous pickup method, but we dropped it after recognising that past behaviour did not reliably predict current intent. The extra few seconds were worth it: a slightly slower checkout was safer than sending customers to the wrong pickup location.
Craft
The interface treated the pickup method as a primary order decision rather than secondary fulfilment information.
The previous flow allowed users to move from order type & restaurant selection into checkout without a strong confirmation of the physical pickup point. The ambiguity survived all the way into the restaurant.
The revised architecture inserts the pickup location after choosing between pickup or delivery. The system cannot continue until the customer explicitly chooses where the hand-off should happen.
As-Is flow [Link to Figma Prototype]
Prototype: New flow [Link to Figma Prototype]
Retrospective
My main regret was not defining event-level analytics before the pilot, which limited how confidently I could isolate cause and effect. The first design also required the same pickup decision every time; a better iteration could remember a previous preference while still requiring confirmation. The broader lesson was simple: when a digital choice affects the physical handoff, the system should never leave that state implicit.
