Safer server operations

To maintain the security of customer data, AWS developers use an internal tool to remotely access and run commands on servers. For hosts with customer data, developers must get secondary approval from a teammate or leader before running a command — a process that can take anywhere from five minutes to several hours. Frustratingly, developers weren't warned an approval was required until after attempting to run a command.

With only three weeks for research and design delivery, I began with a review of previous customer feedback combined with a heuristic evaluation of the workflow. I identified three key pain points: 1) error messages when approval was required were long and confusing, often containing 15+ lines with complicated conditional logic; 2) the UI presented two approval options simultaneously without making it clear that only one was needed; and 3) once an approval request was initiated in the background, there was no indication in the UI when its status changed.

My solution addressed each pain point directly: 1) triggering a warning before execution on hosts requiring approval so developers weren't caught off guard; 2) utilizing progressive disclosure to surface only the relevant approval method; and 3) using polling to keep the approval status updated in real time. I also introduced a notification for developers who navigated away, informing them when their approval was granted and execution completed.

Combined with supporting backend changes, this led to a 4% CSAT improvement, with customer feedback that the simplified flow was easier to understand.