Story Point Inflation
Story point inflation is the upward drift of estimates over time, so velocity rises without more real work. It destroys forecasting and trust. Stop treating velocity as a target, re-anchor to a stable reference, and prefer throughput and cycle-time metrics.
What It Is
Story point inflation occurs when the number of points a team assigns to work of a given size creeps upward over time. A task that was a 3 last quarter is now a 5, not because it is harder, but because points have quietly devalued. Velocity climbs on paper while actual delivered value stays flat. The points become a currency that loses purchasing power.
This defeats the purpose of relative estimation, which exists to help a team forecast, not to produce an impressive-looking number.
Why It Happens
- Velocity as a target. When management treats rising velocity as a goal, teams oblige by inflating estimates — a textbook case of Goodhart's Law.
- Cross-team comparison. Comparing teams by velocity pressures each to inflate.
- Estimate padding. Teams add buffer to protect against overruns, and the buffer becomes the baseline.
- Estimation drift. Without a stable reference task, the team's sense of "a 3" slowly shifts.
Why It Hurts
Velocity becomes meaningless as a forecasting tool — the one job it was supposed to do. Stakeholders relying on velocity for planning are misled. The metric is gamed rather than informative, and once one number is gamed, trust in estimation erodes generally. Energy goes into estimation politics instead of delivery. Worst of all, the team can convince itself it is accelerating when it is not.
Warning Signs
- Velocity rises steadily, but the amount of working software shipped does not.
- Points are treated as targets or used to compare teams.
- Estimates feel padded "to be safe."
- Discussions focus on hitting a velocity number rather than on the work.
Better Alternatives
- Throughput metrics. Count completed items or measure cycle time, which are harder to inflate.
- Reference-class forecasting. Anchor estimates to historical actuals rather than abstract points.
- Relative-estimation discipline. Keep a stable reference story so a 3 stays a 3.
- No-estimates approaches. For mature teams, slice work to similar sizes and forecast by count.
How to Refactor Out of It
First, stop using velocity as a target or a cross-team comparison; that single change removes most of the incentive to inflate. Re-anchor estimation to a stable reference story the whole team agrees on, and revisit it periodically. Consider shifting to throughput and cycle-time metrics, which resist gaming and give better forecasts. Use historical actuals, not aspiration, to predict delivery. The point of estimation is honest planning, so protect it from becoming a scoreboard.