Estimate Range & Confidence Forecaster
Replace fake precision with ranges, assumptions, and risk buffers people can actually trust
Estimation is inherently uncertain, especially for novel work, yet "precise" estimates get treated like deadlines. This recipe forces discipline: break down work, estimate ranges, document assumptions, and attach confidence levels and buffer logic.
PROMPT
Create a skill called "Estimate Range & Confidence Forecaster". When I provide a work description, you will: - Decompose it into chunks and identify unknowns explicitly. - Produce estimate ranges (low/likely/high) and label confidence (1–5). - List assumptions and reforecast triggers. - Provide a stakeholder-ready estimate statement that emphasizes uncertainty and ranges. - Recommend buffer policy (and explain why), unless I disable buffers. Guardrails: - Never output a single "perfect" date unless I explicitly request it. - If critical unknowns exist, propose discovery tasks and gate the estimate on them.
How It Works
Describe the work (epic, feature, deliverable) and any known constraints. The recipe
decomposes it into estimatable chunks, identifies unknowns, produces three-point ranges
(optimistic / most likely / pessimistic) with confidence ratings, and generates a
stakeholder-ready estimate statement that emphasizes uncertainty — not a single "perfect" date.
What You Get
- Work breakdown: decomposed chunks with unknowns called out explicitly
- Estimate ranges: low / likely / high per chunk, with confidence (1–5)
- Assumptions list: everything the estimate depends on
- Buffer recommendation: risk-based buffer with rationale
- Reforecast triggers: conditions that invalidate the estimate (assumption broken, dependency slips, scope change)
- Stakeholder-ready statement: ranges and uncertainty, not false precision
Setup Steps
- Describe the work: objective, constraints, team size, dependencies
- Provide any historical reference points (similar past work)
- List known risks and unknowns (or let the recipe elicit them)
- Review the estimate pack and send the stakeholder statement
Tips
- If critical unknowns exist, the recipe proposes discovery tasks and gates the estimate on them
- Estimation units: hours, days, story points, or T-shirt sizes
- Range models: 3-point or simple low/high
- Buffer policies: risk percentage, fixed days, or none
- Great for rewriting estimate emails so they don't sound like hard deadlines