Coupa spend error tail supplier count changes after hierarchy update

Fixes a Coupa spend analytics error where the tail supplier count changes after a supplier hierarchy update. Use when tail spend counts shift right after a hierarchy change with no new transactions. Not for genuine new supplier onboarding.

Freeze the comparison. Run the tail supplier report for the same period and the same spend threshold twice, noting the count before and after the hierarchy update. Expected: only the count and grouping changed, total spend for the period is unchanged.

Export the supplier lists from both runs and diff them. Expected: the difference is a named set of suppliers that moved, not a broad unexplained change.

Check each moved supplier in the supplier hierarchy. Expected: its parent or rollup node changed during the update, which explains why it entered or left the tail count.

Decide how the tail report should treat parent rollups: count suppliers at leaf level or at parent level, and confirm that choice with the spend analytics owner. Expected: one stated counting rule.

Rebuild or refresh the hierarchy mapping used by the report, then re-run the report for the original period. Expected: the tail count is stable across a second run with no hierarchy changes in between.

Annotate the report for the update date so readers do not read the count jump as procurement behaviour. Expected: the published report carries a note naming the hierarchy update as the cause.

Resolved from the public problem: https://vectle.com/posts/pst_rBlH-M09lFjmRfESqdsEoQ

Problem query: Coupa spend error tail supplier count changes after hierarchy update

TL;DR: A hierarchy update regroups suppliers, so suppliers that were counted under a parent can move into or out of the tail bucket without any spend changing. Re-run the tail analysis against the same date range before and after the update, check whether the moved suppliers changed parent, and rebuild the hierarchy mapping used by the report. If the same suppliers appear under a new parent with identical spend, the count change is a mapping effect, not new tail spend.