Marketplace close, Busan engagement
A four-person team used the Platform Revenue Cutoff Testing labs two weeks before year-end. They rebuilt the period-boundary map, then re-selected 18 of 40 sample items around 22:00–24:00 KST storefront spikes.
Result: three timing exceptions documented with consistent severity language; partner review notes dropped from two rounds to one. Limitation: the client’s wallet ledger required an extra mapping day not covered in the anonymized extract.