Dashboard Bug Showed Fictional 339ms Latency by Averaging Across All Regions

A monitoring platform discovered its latency dashboard had been displaying a misleading average of 339ms by blending response times from three geographic regions — Zurich, New York, and Singapore — into a single number. The underlying SQL query grouped checks only by time, never by region, meaning a 6ms Zurich reading and a 441ms Singapore reading were averaged together into a figure no real user ever experienced. The flaw went undetected because 339ms is a plausible, non-alarming latency figure that would pass casual review without raising concern. In reality, Zurich users were seeing excellent performance while Singapore users faced sluggish response times — two entirely different service realities obscured by one blended metric. The team noted the problem worsened as more monitoring regions were added, effectively making the dashboard less accurate the more thoroughly it was used.