DoorDash is preparing a new fraud detection rules engine for its merchant payouts product after a recent spike in chargebacks increased monthly losses by 18%. You are the program manager leading a cross-functional team of 9 people (3 engineers, 2 data scientists, 1 analyst, 1 product manager, 1 risk operations lead, 1 QA lead) to ship an initial release before the holiday order surge in 8 weeks.
The tension is clear: the data science team wants two more weeks to improve model calibration and reduce false positives, while Finance and Operations want a faster release to stop payout fraud immediately. Leadership has asked for a recommendation on whether to launch a simpler rules-based version now, delay for a more accurate version, or phase the rollout.
The VP of Risk wants fraud loss reduction this quarter. The Head of Merchant Operations is concerned that false positives will freeze legitimate merchant payouts and increase support tickets. The Engineering Manager wants to avoid a rushed launch that creates production instability. Finance expects at least $1.2M in annualized loss prevention if the release goes live before the holiday peak.