**Freeze the comparison basis.** In the spend report, note the hierarchy version, the spend period, and the tail definition in use (spend threshold or percentage of total spend). Expected: both the before and after runs use the same period and the same tail definition; only the hierarchy version differs.

**Diff the hierarchy versions.** Open the supplier hierarchy in Coupa and compare the published version against the previous one, focusing on re-parented suppliers and merged nodes. Expected: a list of suppliers whose parent node changed; these are the suppliers that moved tail bands.

**Check for duplicates and inactive suppliers.** Search the moved suppliers for duplicate records and suppliers marked inactive that still carry spend in the period. Expected: duplicates are merged into the surviving record and inactive suppliers with no spend are excluded from the count.

**Verify the tail threshold at the new node level.** A supplier that was above the tail threshold at a large parent node can fall below it at a smaller child node, because the threshold share is calculated per node. Expected: the count difference roughly matches the number of re-parented suppliers near the threshold.

**Refresh the rollup and re-run.** After the cleanup, refresh the spend analytics data and re-run the report on the pinned basis. Success check: the tail count is stable across two consecutive refreshes and total spend still matches the original figure.