Why Fibonacci-style estimates widen with uncertainty

The reasoning behind the scale

Small items can often be compared closely. For larger items, unknowns and dependencies make fine distinctions less credible. A Fibonacci-style scale reflects that reality by offering fewer choices as estimates increase. Teams may use the exact sequence or a modified version; consistency and shared meaning matter more than mathematical purity.

A vote of thirteen is not a prediction of thirteen hours or days. It says the item looks substantially larger or less certain than the team’s smaller references. If several people choose values far apart, the spread is an invitation to explore assumptions.

How to run a round

1. Present one prepared work item and its expected outcome.
2. Answer clarifying questions without suggesting a target estimate.
3. Ask every active estimator to choose independently.
4. Reveal all choices simultaneously.
5. Invite the highest and lowest estimators to explain what they considered.
6. Update the item or assumptions, then vote again if needed.

Observers can help clarify product or technical context without being required to vote. The facilitator should create space for every estimator and keep discussion focused on the work.

When a large value is useful

A large estimate can mean more effort, more coordination, more technical uncertainty, or all three. Before accepting it, ask whether the item has a clear outcome and whether it can be split into independently useful slices. Splitting should improve understanding or delivery flow, not merely force every item below an arbitrary point threshold.

Fibonacci versus other scales

T-shirt sizes are approachable for early, coarse comparison. Linear numbers may suit highly repeatable work but can invite debates about differences that the team cannot reliably perceive. Custom decks can represent confidence, risk, or specialized operational units. Whatever you choose, define what the values mean and avoid switching scales midstream without a clear reason.

Improve the conversation

• Keep reference items visible during estimation.
• Do not average votes automatically before discussing disagreement.
• Separate value and priority decisions from size.
• Review unexpectedly difficult items in retrospectives to improve shared understanding.

For more context, read about story points and the complete agile estimation workflow.