Estimation & Sizing

Help a team size work honestly and translate sizes into a range, not a fake-precise date.

When to use

• Sizing a set of stories/initiatives, or producing a delivery forecast from scope.
• Sanity-checking an estimate that feels too confident.

Principles

• Estimate relative size (T-shirt: XS/S/M/L/XL, or points) before absolute time.
• Always estimate a range (optimistic / likely / pessimistic), never a single number.
• Make assumptions explicit — every estimate rides on them.
• Bigger = less certain. Break L/XL items down before committing.
• Account for the non-coding tax: review, testing, integration, meetings, ramp-up.

Process

1. Clarify scope per item; flag anything too vague to size (size = "needs breakdown").
2. Assign a relative size and a confidence (H/M/L). Split low-confidence big items.
3. Capture the assumptions behind each estimate.
4. Convert to time using a stated velocity / throughput or size→days mapping — show the conversion.
5. Roll up to a range and add buffer for risk/unknowns proportional to uncertainty.
6. Present as "likely X, could be Y–Z" with the drivers of the spread.

Output format

| Item | Size | Confidence | Assumptions | Est. range |
| --- | --- | --- | --- | --- |

**Rollup:** likely {{X}}, range {{Y–Z}} ({{unit}})
**Key drivers of uncertainty:** {{...}}
**Assumptions:** {{...}}

Resist pressure for a single date — offer the range and the conditions under which the optimistic end holds.