You are the engineering manager leading a checkout rewrite for a B2B commerce platform that processes a meaningful share of annual revenue during a six-week peak season. The legacy checkout has reliability issues and blocks several contracted customer commitments, but the new version is only partially complete and the highest-risk work sits in payment retries, tax calculation edge cases, and rollback automation. The commercial team is pushing hard to launch before peak season because several enterprise accounts were sold on the new flow, while your principal engineer wants to delay until all known defects are resolved after a recent staging incident exposed data consistency gaps. At the same time, one upstream dependency on identity changes is slipping by two weeks, and you have no approval for additional headcount or a full parallel run.
| Detail | Value |
|---|---|
| Engineers assigned | 6 backend, 2 frontend, 2 QA |
| Traffic during peak | 4.5x normal weekly order volume |
| Deadline | 10 weeks before peak season starts |
| Current known Sev-2 defects | 11 |
| Enterprise customers waiting | 5 signed accounts |
| Allowed checkout downtime | < 15 minutes total during cutover |
| Budget for external support | $60K |
| Compliance requirement | No tax or payment reconciliation errors in production |
How would you decide the right trade-off between speed and quality here, and how would you execute the launch plan so the business can move quickly without taking unacceptable customer, revenue, or operational risk?