What consistent bias tells you that any single story can't
One story running over its estimate could be anything - a fluke, hidden complexity, a bad day. A consistent pattern across many stories - the team is reliably estimating 30% under actual effort, say - is a calibration signal, and it's the kind of signal that only shows up once you're tracking more than one data point.
What to do with a measured bias
A team that discovers it's systematically underestimating doesn't need to abandon story points - it needs to recalibrate its reference stories, the same process covered in calibrating your estimation scale. The bias itself is useful information, not evidence the whole practice is broken.
Privacy
Everything you enter here stays in your browser's local storage - nothing is sent to a server, and there's no account or sign-up. Clearing your browser data clears this tool's history too.