ReimburseMe

A pharmacy technician covering shifts at three different stores in a week has to answer a question the system never made easy: which store pays for what. Until 2024, the answer came on paper, through a shared computer in a back room. Associates had to wait hours each week for validation before they could even submit an expense. That time often came out of the technician's own evenings.

Expense reporting was fragmented across paper forms, legacy tools, shared computers, and business-specific policies. For associates splitting time across multiple locations, the process was error-prone and paper-dependent. A validation step cost hours every week before a report could even be filed. Managers had to navigate approval rules that differed by group and by market.

The two ends of this problem pulled hard against each other. Store and field associates often had no personal computer, and sometimes no digital workflow at all. At the same time, every report had to satisfy country-specific tax rules, financial controls, audit requirements, and approval policy.

I joined after the initial release, when the product needed more functionality and a mobile experience. Early on, I extended the information architecture, found gaps in requested flows, designed responsive experiences, built interactive prototypes, and tested them with associates.

As the product expanded, my role expanded with it. I became the primary design resource for enterprise adoption. I led design planning for the Walmart US Stores, Sam's Club, and Supply Chain launches, worked through implementation with product and engineering, and onboarded an associate designer as the scope grew beyond what one person could handle.

The expense process presented itself as a collection of separate forms and policy requirements. I reframed it around a common task model: start a report → add eligible expenses → resolve issues → submit → review or approve.

That shared structure let us keep one consistent experience and add policy-specific rules only where they mattered. Instead of building a different product for every associate group, we built one common foundation. It could handle differences in forms, receipts, approvals, attendees, delegation, and reimbursement rules.

Before development handoff, we ran moderated 30–45 minute usability sessions with 35 participants: store associates, managers, and market directors of Health and Wellness across Walmart and Sam's Club. The sessions covered 10 flows across the core reporting and approval workflows.

87% task completion

Most flows performed well. Policy switching and part of the approval experience did not. The cause turned out to be language, not structure.

Associates didn't recognize the word "policy." In stores, the real idea is a charging store or charging department: which location pays for it. For a technician working three stores in a week, that distinction is the transaction, and they already had their own words for it. We had named the feature after our internal model of the system, and asked people to learn it just to use it. It became Change department.

Approvals had the same problem, twice over. "Tolerances" is finance vocabulary. Managers read it as jargon and skipped past it, so a control we designed to be visible wasn't actually working. It became View expense limits.

"Escalate" failed in a different way, and the difference mattered. People recognized the word, but they didn't know what escalating a report would actually do. So a better label wouldn't have helped. That one needed an explanation, not a rename. We rewrote the copy around the confirmation step itself, in the modal on desktop and the equivalent bottom sheet on mobile, so associates saw what escalating meant before they committed to it.

We shipped the revised copy without a second round of testing. The handoff date was fixed, and a re-test would have cost more than it would have told us. The changes were just terminology corrections, taken directly from the words associates used in the sessions. They weren't new design ideas that needed testing.

The roughly $20M comes from three sources: fees avoided by retiring a per-report SaaS tool, automation like auto-approval, and associate time we gave back. Part of that is booked savings. The larger part is an internal estimate of recovered associate time, and I'd present it as an estimate, not a booked saving. That distinction matters when finance is in the room. Showing that you know the difference matters even more.

By FY26, ReimburseMe was handling over 900,000 reports a year across multiple business groups, on a common platform that could take on new groups and markets without rebuilding from scratch.

Three enterprise launches in 13 months

Those results came from three launches I led design planning for: Walmart US Stores (October 2024), Sam's Club Stores (summer 2025), and Supply Chain Facilities (November 2025). For each, I worked with product and engineering to understand the operating context, identify gaps, determine what could stay shared, and plan the design work to release.

Then two countries launched on the same design

Mexico (March 2026) and Chile (August 2026) launched on the same design. Another designer led them. I supported as designer and platform SME. The remaining work was country-specific edge cases, not new architecture. The shared task model held across borders. That's the clearest evidence the foundation worked.

The patterns outlived the product

The reach didn't stop at ReimburseMe's own markets. Patterns from this project (the shared task spine, the policy-switching model, the receipts-plus-OCR approach) carried into the finance platforms around it. I designed GNFR Invoicing, including its Mexico implementation, and reused them there myself. AP Imports and Claims Reimbursement were not mine. I didn't lead those projects, and I'm not claiming credit for their results. What traveled was the thinking. Teams solving related problems reused the structure instead of reinventing it. Influencing work across an org you don't control is a different, harder kind of scale than shipping your own screens. It's the part of this project I'd point to first.

As scope expanded, my responsibilities moved past individual flows. I coordinated design priorities across releases, kept product and engineering aligned on scope, and supported launch planning.

I also mentored an associate designer, from Figma fundamentals to owning ReimburseMe features end to end: discovery through handoff, including contributions to the platform's component library. I stayed hands-on with the product, while building enough clarity that someone else could build on it.