Our biggest releases used to be first come, first served. We had run request-based releases before, but nothing at this scale. I owned the move to requests: Members choose the trips they want and rank them, and the system allocates across everyone who entered, using ranking and multi-award logic I defined with the business and engineering.
The build ran three months with more than ten business stakeholders, including the CMO, CTO and President, and directives that kept changing. My job was holding scope and protecting testing time through all of it.
Less than three weeks before the release went out to 4,200 members, the President and CMO raised concerns about what had been built. I prototyped an alternative over a weekend, met the dev team on a day the office was closed to test whether it would work, got the same stakeholders aligned, and had the new approach running in production less than two weeks later.
The release drew 17,876 requests, 7.8 times the ~2,300 of the biggest release before it. 1,816 unique memberships entered, up from a high of 650, for travel across 14 months and 72 destinations. Members added about 40% of the requests themselves and their Account Managers added the rest, all through one request flow.