In 30 seconds
- At the idea stage, estimates can be off by about 4x either way.
- Uncertainty narrows only as real decisions are made and work is done.
- Give ranges early; tighten them as you learn.
What it shows
Plot estimate accuracy against project progress and you get a cone. At the very start, a project could take a quarter of the estimate or four times it. As requirements, design and building progress, the range narrows.
Where it comes from
Barry Boehm described the pattern in his 1981 work on software cost estimation. Steve McConnell named and popularized the cone in his books on software estimation.
How agile teams use it
- Give early forecasts as ranges, never single dates
- Re-forecast every sprint using real velocity
- Reduce uncertainty early with spikes and thin end-to-end slices
See Forecasting.
Myth vs reality
Myth · Better estimators beat the cone.
Reality · The uncertainty is in the problem, not the estimator.
Myth · The cone narrows by itself over time.
Reality · It narrows only as decisions are made and risks are resolved.
Myth · Agile removes the cone.
Reality · Agile works within it by learning faster.
FAQ
How wide is the cone at the start?
Commonly cited as 0.25x to 4x of the final result.
Can a project's uncertainty grow?
Yes, if scope changes or key decisions are reopened.
How do I explain this to stakeholders?
Show a range that narrows each sprint as evidence comes in.