Designing an exit flow that gives people a reason to stay

Context

Before this project, Amazon didn’t have a single self-service way for employees to resign, see what would happen next, or find resources that might address why they were leaving. I was the lead designer and owned the full experience in the A to Z app, on mobile and web. I worked with Engineering, Product, Legal, Compliance, and People Experience leadership from early strategy through launch.

On paper, it was a multi-step form with some conditional logic. The hard part was everything around it, and honestly, the most useful work I did happened before we designed a single screen.

Challenge 1: Five teams, five definitions of success

Every group had a real stake in this. Employees who were already stressed needed empathy, not paperwork. Legal and Compliance needed exact language, documented consent, and reliable data capture. HR leadership wanted retention options in the flow without it feeling manipulative. Product and Engineering needed something that would scale globally.

What worked: aligning before designing

I started by meeting with People Experience leadership to agree on a shared vision. That included retention metrics we could actually measure and a clear picture of how the experience would affect the people going through it. Then I ran workshops with stakeholders from all five groups.

Doing this early made a big difference. The workshops surfaced assumptions that would otherwise have turned into arguments in design reviews, and we came out with a shared framework we kept going back to whenever we had to make a tradeoff.

Challenge 2: Two populations, two legal frameworks

The flow had to work for full-time employees and contingent workers, who have different access policies and legal protections. If we’d designed for one group and adapted it for the other, it would have broken in ways that were hard to spot.

What worked: side-by-side service blueprints

We interviewed current employees who had thought about resigning and former employees who had actually left. From those interviews, we built service blueprints that tracked both groups side by side. That showed us where their needs were different and where one design could cover both. People Experience teams have since reused that approach on other projects.

Challenge 3: Offering alternatives without pressuring people

HR wanted retention options in the flow. Employees needed to feel respected, not sold to. If we got that wrong, the whole thing would feel like a trap.

What worked: let the reason decide what people see

The flow starts with a short personal acknowledgment, then asks for the main reason someone is leaving, with follow-up questions specific to that reason. Everything after that depends on the answer. If someone’s leaving for growth, they see training resources, career development materials, and a link to the A to Z Assistant. If it’s about scheduling, they see alternate shift options.

We were also careful to protect people’s right to choose. The last step shows a full summary they can edit before submitting, and it’s clear they can change their mind up to 24 hours before their last day. The cancel modal adds a little friction to confirm intent, but not enough to get in anyone’s way.

Challenge 4: Making the case for an investment in people

The hardest part was justifying the project at all. It didn’t make money, so proving its value meant translating things like retention and satisfaction into dollars.

What worked: putting a number on attrition

I estimated what it costs to replace someone who leaves, including recruiting, onboarding time, and lost productivity, and compared that to the retention lift from the flow. That’s what got us long-term support, and it changed how People Experience leadership looked at employee-facing projects from then on.

What changed halfway through

Partway through, research and stakeholder conversations turned up something we hadn’t expected. Employees didn’t just want more training. They wanted access to training from other teams across Amazon. A lot of them weren’t looking to get better at their current job. They wanted to see where else they could go inside the company.

That changed what we were building. The resignation form became one part of a bigger retention effort, centered on an Explore hub that brought together career growth resources, team openings and leader profiles, flex shift options, and self-service resignation. Each one gave people another possible reason to stay.

Results

The flow measurably reduced attrition in the groups it served, moved NPS in the right direction, and cut recruiting spend, both in our model and in practice. The way we connected a structured form to the A to Z Assistant is now used in other self-service HR features.