Most cutoff findings that unravel in partner review share one habit: the team selected samples before they agreed what “the period” meant for the platform. Storefront clocks, settlement files, and recognition policies rarely share a single midnight.
Start with three clocks
Write down the storefront timezone, the settlement batch timezone, and the recognition policy timezone. If any two disagree, draw them on one page. App Platforms learners reuse the boundary map from Module 2 for this step; the diagram matters more than the prose.
Pull the shipping or fulfillment log next
For physical goods marketplaces, fulfillment timestamps often trail payment authorization. For digital goods, “fulfillment” may be entitlement grants. Either way, place a 48-hour window around the cut date and mark every transaction whose clocks disagree.
Only then design the sample
Once the boundary map exists, stratified selection around late-day spikes becomes obvious. Skipping the map usually produces neat-looking random samples that miss the risk.
Document residual judgment
Partners calm down when they see what you did not test and why. A short residual risk note — three sentences — often prevents a second review round.