Platform storefronts in Korea often spike between 20:00 and 24:00 KST. Monetary unit sampling (MUS) still works mathematically, but if high-value items concentrate earlier in the day, late-day timing errors can be underweighted.
Stratify on time and SKU velocity
A simple two-way stratification — hour band crossed with top SKU cohort — usually beats a larger flat MUS sample for cutoff objectives. You are testing timing, not only existence of value.
Keep the performance materiality honest
Do not shrink sample sizes because the platform “looks automated.” Automation moves the risk to configuration and batch timing, which still needs evidence.
Lab habit
Module 4 of our flagship course includes a stratification table you can paste into Excel or your firm’s sampling tool. Adapt the hour bands to the client’s local peak, not to ours.