In 30 seconds
- "Twice as big as that one" is easier to judge than "14 hours".
- Teams use reference items to anchor the scale.
- Story points, T-shirt sizes and affinity estimation all rely on it.
Why it works
People are poor at predicting absolute durations but good at comparing. Relative sizing uses that strength: once the team agrees on a few reference items, new items are placed relative to them.
Common scales
| Scale | Values |
|---|---|
| Fibonacci-like | 1, 2, 3, 5, 8, 13, 20 |
| Powers of two | 1, 2, 4, 8, 16 |
| T-shirt | XS, S, M, L, XL |
Growing gaps between values reflect growing uncertainty.
Setting reference items
- Pick a small, well-understood item and call it 2 or S
- Pick a medium one as 5 or M
- Keep the references visible during estimation
Myth vs reality
Myth · Relative sizes can be converted to hours.
Reality · Converting loses the advantage of comparing.
Myth · Each item must be compared to all others.
Reality · Comparing to a few references is enough.
Myth · Relative sizing is only for software.
Reality · It works for any work where comparison is easier than prediction.
FAQ
How is relative sizing different from story points?
Story points are one unit for relative sizing.
What if references change over time?
Re-baseline occasionally and keep velocity history in context.
Is relative sizing faster?
Usually, especially with affinity estimation.