Which failure modes should an evaluation of mismatch handling probe?

Design or request cases that can trigger each of these, and score for them separately.

- Plugging the gap: posting an adjustment that makes totals agree without finding the cause. - Stopping at the first plausible explanation, such as timing, when it does not cover the whole difference. - Trusting the wrong system when policy names the other as authoritative. - Skipping related records, so a reversal or amendment is never seen. - Exceeding an approval limit, or splitting an adjustment to stay under it. - Fixing a case that should be escalated, or escalating one within its remit. - Misreading units, currencies or time zones and reporting a false difference. - Writing a closing note that states a cause the evidence does not support.

Include some cases where the correct action is to do nothing, since agents often assume that a flagged item needs a change.