Bay Decontamination
One task, two teams, one experience
Bay Decontamination takes up about 30% of General Service time at The Home Depot, roughly 6 million labor hours. I redesigned it as a microservice in BOLT, so MET associates could scan SKUs, see where they belong, and finish each bay without switching apps. Along the way, I found that Store Ops was about to build the same experience for Sidekick, and I aligned both organizations on building one reusable component.
Project Overview
Role: Senior UX Designer, lead designer on Bay Decontamination. I owned design, usability testing, and synthesis, and created the associate rollout materials.
Team: MET Product, MET Labor & Process, MET Engineering, and me. I partnered with Store Ops Product, UX, and Engineering to align on one shared component.
Timeline: Research, design, and testing through April 2024
Platforms: BOLT (MET associates' task app) and Sidekick (Store Ops and Freight associates' task app)
Users: MET Associates (MEAs) and MET Supervisors, plus Freight associates and night operations leaders
Methods: Contextual inquiry (in-store observation and interviews), prototype usability testing in two rounds, UMUX-LITE, synthesis in Miro, cross-team alignment readout
Problem Statement
MET associates (MEAs) service bays for vendors who want every bay "grand opening ready." Decontamination is a required part of that work: front-facing products, matching each model to its price label, removing misplaced SKUs, and clearing trash.
1. Too many clicks to find a SKU. To check a SKU, MEAs left BOLT, opened Inquiry, and clicked through to the Digital POG. Switching between apps slowed every bay.
2. A required task with no visibility. Decontamination was a single Yes/No checkbox with no task-level time tracking. The business couldn't see where the time went.
3. Everyone had their own method. Without guidance in the app, each MEA developed a personal way of working, often with slower manual steps.
How might we
let MEAs scan SKUs directly in BOLT and show them only what they need to finish the bay quickly?
My Role
I was the lead designer on Bay Decontamination, and I owned the work from store research through the associate rollout materials.
-
Research: led store observation and both rounds of prototype testing in Atlanta and NYC, and synthesized the findings in Miro.
-
Design: designed the decontamination microservice in BOLT, including the SKU scan, the compact product card, the Digital POG view and the task flow.
-
Strategy: raised the gap between the business's time-tracking plan and how MEAs actually work, and brought the task-order findings to the Labor team.
-
Alignment: compared MET's and Store Ops' decontamination checklists, and led the readout that aligned both organizations on one reusable component.
-
Rollout: created the associate one-pager and the V2 user guide for the new microservice.
The microservice rolled out to all stores.
Research
Before changing the screen, I needed to see how decontamination actually happens in the aisle.
I observed MEAs (MET Associates) during General Service and tested prototypes in stores in Atlanta and NYC. The first round included 16 MEAs and 7 supervisors. By the end, 31 people had taken part: 20 MEAs, 9 MET supervisors, a freight associate, and a night operations Assistant Store Manager. I synthesized everything in Miro.
Synthesis in Miro: store visits, two rounds of testing, and the notes that became the five key insights.
Watching the work, not just the app. The store floor showed pressures the task list didn't account for:
-
Stores filled empty bay space with items that didn't belong, adding work before decontamination could start.
-
Understaffed stores meant customers pulled MEAs away, sometimes for more than 10 minutes.
-
Some bays had very complex POGs. "There are 69 pages of POG for this bay alone." (MEA)
-
The app sometimes froze, and store Wi-Fi was unreliable.
Key Insights
MET associates (MEAs) didn't need more instructions. They needed the right task, in the right order, with the right information in hand.
Strategy: Business vision VS the Floor
The business wanted to time one task. On the floor, MEAs mixed it with several others.
The business wanted to track time spent on decontamination alone. In the stores, I saw MEAs mix it with setting the bay to POG, front-facing, handling damaged SKUs, scanning lows and outs, and packing down, often without checking the phone until the bay was done.
I raised this gap with the business and asked: Are we simplifying or complicating the workflow? The business and Labor teams then set the direction:
-
Set Bay to POG comes first, with decontamination second or third.
-
Each bay gets only the tasks it needs, and each task is timed.
-
MEAs put misplaced SKUs in a cart for supervisors instead of returning each one, and they don't make markdown decisions on damaged products.
My job was to design a flow that worked within those decisions: time is tracked in the background from start to end, the next task opens on its own, and the pink banner sends misplaced SKUs to the cart.
Design Principles & Decisions
Research surfaced challenges beyond usability. They became four principles, and every design decision traces back to one of them.
1. Keep associates in flow
Principle: Associates switched between tools to find what they needed for one task. Bring the critical information into one workflow so they stay focused.
-
Scan in BOLT. MEAs scan or type a SKU right inside the task. A compact product card shows price, description, photo and location, so there's no trip to Inquiry or SKU Depot. Scanning is optional, so it helps without adding a step.
-
The next task opens on its own. MEAs had asked not to keep clicking to open each card.
2. Make effort visible without friction
Principle: A single Yes/No hid how much effort each bay took. Capture effort inside the workflow without adding work for associates.
-
Time tracked in the background. MEAs never see a timer. Only supervisors see task time, so associates focus on doing the bay right instead of rushing it.
-
A check before "done." An "End Task?" check keeps MEAs from ending the task too early.
3. Design for approximation, not perfection
Principle: Some bays had several POGs spanning dozens of pages. Point associates to an approximate location instead.
-
Location on the Digital POG. The product card highlights the SKU's spot in the bay. "If you can give us an approximate location in the bay, that would be fine." (MEA)
-
A color for every next step. A green, yellow, or pink alert tells associates whether the SKU stays, goes back in the aisle, or goes in the cart.
4. Build once, scale across teams
Principle: MET and Store Ops (including Freight) did nearly identical decontamination in separate tools. Create one reusable experience.
-
One decontamination component for both organizations' apps. How the two organizations got there is in build once, use twice below.
Testing and Iteration
UMUX-LITE rose from 83.37% to 92.3% after iterating, and every tester responded positively to scanning in the task app.
What we left out
-
Extra questions at the end. 10 of 18 testers didn't want anything slowing them down, and the data platform couldn't record free text. We kept one Yes/No, "Is the bay fully decontaminated?" (No sends MEAs back to the bay), and planned pre-set reason codes for later.
-
A Pause button. Supervisors worried it would make time tracking less accurate, and only one other store app had tried one. We held off until there was more evidence.
Build Once, Use Twice
aligning MET and Store Ops
Store Ops was about to build the same experience in its own app, so I brought both organizations together to build it once.
-
Compared both teams' standards line by line. They were "so similar that the word identical comes to mind."
-
Led an alignment readout with stakeholders from both organizations.
-
Outcome: both agreed to build one shared decontamination experience.
Rollout and Impact
The microservice rolled out to every store, and MET and Store Ops aligned on one shared component.
-
I created the associate one-pager and user guide for the rollout.
-
Labor figures are projections from MET's labor engineering team. Experience scores come from my usability testing.
Reflection
The biggest lesson: the most valuable design work happened before the screens, in questioning the ask and connecting two teams.
-
The floor beats the brief. The business asked for a timer. Watching MEAs showed how the work really happens, and raising that gap early helped shape what the timer should measure.
-
Look sideways, not just down. The biggest efficiency gain wasn't a feature. It was noticing that another organization was solving the same problem and getting both to build it once.
-
A good "no" is a design decision. Keeping the end question simple and holding off on Pause kept the task fast and the time data trustworthy.
What I'd do next. I'd measure actual decontamination time against the labor projection, design for slow or dropped connections from the start, and add a short pre-set issue list once V1 data shows the most common problems in the bay.
