Case 05 · Amazon Robotics · ToteASRS Operational Research
Before Amazon Robotics could design new tools, we needed a ground truth of how an entire fulfillment center actually runs.
UX Lead — co-led with partner UX
Product, ToteASRS, UX
March 2023
Hundreds of metrics are monitored constantly by different users across different tools and dashboards. The challenge was mapping all of it before designing anything new.
An average Amazon Robotics fulfillment center has ~1,300 direct-labor associates and 160 operations leaders and support staff, working two shifts a day across four periods each.
Five organizational levels — GM, Sr. OM, OM, Area Manager, and PA (Process Assistant) / QB (Quarterback) / AFM (Amnesty Floor Monitor) — each with different tools, metrics, and decision rhythms.
Hundreds of metrics across dozens of tools owned by Amazon Robotics, Amazon Fulfillment Technologies, and other internal orgs. Operators were the only people connecting them.
I led the UX side of a week-long ground-truth research effort at a Houston fulfillment center, cross-referenced against two additional FCs across the US, to map the full operational experience as input to a cross-org Tools North Star.
One week onsite, three FCs in scope, every level of the org represented. The output wasn't a survey — it was a side-by-side map of which metric or data point each role pulls from which dashboard, on which screen, at which moment of the shift.
I mapped the five-level operational structure to the actions, tools, and decision cadences each role depends on.
Owns the FC's KPIs, profit and loss, and network-level performance.
Leads a process path (Inbound, Outbound, CAP, Change Management).
Keeps a single process path moving — staffs, trains, and leads Area Managers.
Owns a specific area: safety, engagement, standard work.
Right-hand staff: SOP compliance, floor health, real-time issue resolution.
Across every Ops role I mapped, daily work was made up of the same five actions — at different time horizons and through different tools. Mapping work as verbs (not job titles) let us compare tooling needs across roles — and exposed where the same task was forcing a manager to switch tools mid-flow.
Senior managers evaluate prior-day performance and periodic targets at shift start.
Mid-level managers watch execution throughout the shift and take proactive action.
Managers plan staffing and adjust labor share across process paths to hit targets.
Support managers identify and resolve issues — staffing, equipment, engagement.
Every role reports performance and issues upward, daily and per shift.
Operators told us repeatedly that the tool ecosystem is the problem, not the metrics themselves.
"TELL ME WHAT TO DO — the tools should prioritize and make decisions for the building. Ops just needs to execute."
"In my first month as OM, I just kept receiving links. The tools change all the time; many of my bookmarked links are out of date."
"In an ideal world, when I start my shift, Vantage and Flow tools should be integrated and give me one overview of how the building is running."
"There are way too many tabs. Some load slowly and crash. I find myself just sitting in front of my laptop waiting."
I turned the research into a cross-organizational vision for a centralized, role-aware, decision-supporting operator tool.
The research re-shaped how Amazon Robotics planned the next generation of its operations tools.
Mapping every role, action, and tool side by side made it impossible to keep treating manager UX as a per-feature problem. The work re-positioned tools and metrics as a portfolio investment.
Get in touch
Currently exploring senior & staff-level UX roles in systems thinking and service design.