A late-night setup yielded misleading agent metrics, underlining the need to verify reported performance data before drawing conclusions.
