Amazon Leadership Principles — Behavioral Stories (STAR)

Delaying a release to protect customer trust

• Amazon LPs — Have Backbone; Disagree and Commit · Earn Trust · Customer Obsession · Insist on the Highest Standards · Are Right, A Lot
• Situation — We were expanding into the Australian market, where schools had compliance requirements needing new attendance functionality. The timeline was extremely tight — Australian schools start their academic year in January and we wanted it ready before onboarding.
• Situation — I was leading a significant part of this. By the last week of December development was complete, but the feature hadn't finished QA and regression testing.
• Task — Pressure to release was high, but I believed shipping an untested compliance feature was a significant risk, especially for new customers seeing our product for the first time. My job was to balance the deadline with a reliable experience.
• Action — I reviewed the remaining QA and identified several important scenarios still unvalidated. Since attendance is business-critical, releasing without testing risked production issues during onboarding.
• Action — Instead of simply disagreeing, I requested a discussion with my manager and engineering director. I acknowledged the deadline's importance, then explained why a short delay would actually reduce business risk.
• Action — I presented three points: (1) attendance wouldn't be heavily used in the first week, giving a buffer; (2) QA still had important validation remaining; (3) a buggy first impression would damage customer trust far more than a few days' delay.
• Action — Rather than an indefinite delay, I recommended finishing QA first, then a controlled rollout. After discussing tradeoffs, both my manager and director agreed with the approach.
• Result — We completed QA, fixed the issues, and gradually enabled the feature. The rollout was smooth with no major production issues; product and support later confirmed schools used it without critical bugs during onboarding.
• What I Learned — Sometimes the fastest decision isn't the best decision. Respectfully challenging a decision with data builds credibility — voice concerns when customers may be impacted, then commit once the team aligns.