Replacing a fragmented, paper-heavy inventory process with an intuitive real-time tool that put accurate stock data and time-tracking capability directly in the hands of associates on the floor.
Client
Walmart
My Role
UX Designer & Researcher
Domain
Retail Ops · Internal Tool · Mobile
Year
2018
Tools
Figma · Sketch · InVision · Miro
Overview
Walmart operates over 4,700 stores in the United States, each managing tens of thousands of SKUs across multiple departments. Accurate, real-time inventory data is the operational backbone of retail at this scale — driving replenishment, pricing, staffing allocation, and customer availability. When that data is wrong, the consequences ripple across the supply chain.
The Pulse project was initiated after internal operational audits revealed that a significant portion of store-level inventory discrepancies originated from manual, paper-based adjustment processes that were slow, error-prone, and impossible to audit. Associates were recording adjustments on paper logs that were reconciled with central systems hours or days later — creating a lag between physical reality and digital inventory that generated out-of-stock errors, customer complaints, and unnecessary replenishment costs.
Working through Cognizant's engagement with Walmart, I led the UX design for the Pulse app — a mobile-first internal tool designed to make inventory adjustment and time tracking immediate, accurate, and effortless for frontline associates.
The Challenge
Internal tools at Walmart's scale carry a unique design constraint: they're used by a workforce with enormous range in technical comfort, often in physically demanding environments — on warehouse floors, in refrigerated sections, during peak-traffic hours. The design had to be as robust for a 20-year veteran as it was for a first-week associate.
Manual, paper-based adjustments — Associates recorded inventory changes on physical log sheets that required manual data entry into the central system by supervisors — typically hours after the fact. By the time the data was current, the store floor had moved on.
No real-time visibility — Department managers had no live view of inventory levels across their section. Stock checks required walking the floor or calling the stockroom — a significant time drain across a workforce measured in the thousands per shift.
Fragmented multi-location tracking — For associates managing inventory across multiple departments or store sections, there was no unified view. Tracking across sections meant multiple log sheets, multiple check-ins, multiple sources of truth — none of which were reliable.
Time-tracking disconnected from task context — Associates logged time separately from the inventory tasks they were performing, making it impossible to understand labor cost per task type or identify operational inefficiencies.
Research
Designing for frontline retail workers required going to the floor. Desk research and stakeholder interviews alone could not surface the physical constraints, time pressures, and informal workarounds that shaped how associates actually performed inventory management. I led an immersive research program rooted in the associate's real working environment.
Spent 3 days embedded in two Walmart stores during peak and off-peak hours — observing inventory adjustment workflows, time-tracking behaviors, and inter-team communication patterns firsthand.
28 interviews across associate roles (floor staff, department leads, stockroom, customer service), tenure levels, and departments — capturing the full spectrum of inventory management responsibilities.
Mapped the end-to-end inventory adjustment process in exhaustive detail — every step, every handoff, every tool used — to identify automation and consolidation opportunities.
Iterative wireframe review workshops with store managers, operations leadership, and IT — aligning on functionality requirements and technical constraints before high-fidelity design work began.
Researched the physical conditions under which the app would be used — gloved hands, bright/dim lighting, noisy environments, moving between locations — all of which directly shaped interaction design decisions.
Detailed task analysis of the 8 most common inventory management scenarios — identifying the minimum viable information and interactions required for each task to be completed accurately on mobile.
Physical Context Insight
Associates frequently used devices while wearing gloves or with one hand occupied holding product. Every interactive element in the final design needed to be operable with a single large touch target — eliminating any small buttons, precision gestures, or two-handed interactions from core workflows.
Cognitive Load Insight
Associates made inventory adjustments while simultaneously managing customer interactions, team communication, and physical tasks. The app needed to support "five-second interactions" — tasks completable in a momentary pause without requiring sustained focus or sequential steps.
Trust Insight
Experienced associates were deeply skeptical of new digital tools — they'd seen systems come and go, and had learned that paper was more reliable than software that was slow, required training, or went offline. The app needed to work faster than paper to earn adoption.
Collaboration Insight
Inventory discrepancies were often identified not by the associate responsible for a section, but by someone passing through. The app needed social and notification features to enable cross-department flagging — not just individual task management.
Design Process
Every design decision in Pulse was tested against a single constraint: would an associate be able to complete this task accurately while standing in a store aisle during a busy shift? This constraint eliminated interface complexity quickly and kept the design focused on the absolute essentials.
Multiple rounds of low-fidelity wireframes reviewed in collaborative sessions with stakeholders and associate representatives before any visual design began.
Prototype testing conducted in actual store environments — not conference rooms — to surface physical and environmental issues invisible in lab settings.
Designed a natural language interaction layer for common tasks — associates could report stock levels or flag discrepancies by typing or speaking naturally rather than navigating menus.
Phased rollout to 3 pilot stores with structured feedback collection — two additional design iteration cycles based on real-world usage before broader deployment.
Core Features Designed
One-Tap Inventory Adjustment — Scanning a SKU instantly surfaced the current inventory record with a single large increment/decrement control and a "Confirm & Sync" action — completing a full inventory adjustment in under 10 seconds, with the change synced to the central system immediately rather than hours later.
Real-Time Multi-Location Dashboard — Department managers could view live stock levels across their full section — and flag locations below threshold directly from the dashboard — without leaving the stockroom or making phone calls.
Integrated Time & Task Tracking — Associates logged time in the context of specific task types — "Stocking," "Inventory Count," "Discrepancy Resolution" — creating the first reliable data set linking labor hours to specific operational tasks at the store level.
Cross-Department Discrepancy Flagging — Any associate could flag a stock discrepancy in any department, with the alert automatically routing to the appropriate department lead — operationalizing the informal peer-spotting behavior observed in research into a structured workflow.
Impact
↓ 76%
Reduction in average inventory adjustment cycle time — from hours (paper-to-system lag) to under 30 seconds (real-time sync)
↑ 29%
Improvement in inventory data accuracy rate across pilot stores — directly reducing out-of-stock errors and unnecessary replenishment orders
94%
Associate adoption rate within 6 weeks of pilot launch — driven by the speed advantage over paper processes that earned frontline trust quickly
Key Learnings
Speed is the feature that earns trust with frontline workers. Associates weren't going to adopt a digital tool out of compliance or enthusiasm — they adopted Pulse because it was measurably faster than the paper alternative. The design had a simple goal: be faster than paper. Everything else was secondary.
In-context testing is non-negotiable for operational tools. Conference room testing of the first prototype showed strong results. In-store testing with the same prototype revealed 11 issues invisible in the lab — glare on the screen, button sizes too small for gloved hands, audio alerts inaudible in the stockroom. Context is not a nice-to-have in operational UX; it's the research.
Internal tools deserve the same design rigor as consumer products. There's a common belief that employees will "use what they're given" regardless of design quality. Pulse proved the opposite — associates had a real alternative (paper), and they chose the digital tool only when the design was genuinely better. UX quality determines adoption, and adoption determines ROI.