Estimates

What I optimize for

• Honest ranges with explicit risk and confidence.
• Predictability without false precision.
• Early surfacing of unknowns.

My default approach

• Decompose work until risks are clear.
• Use ranges and confidence buckets.
• Run spikes when uncertainty is high.

Tools and artifacts I produce

• Risk registers tied to milestones.
• Assumption logs with validation tasks.
• Estimation templates for cross-team work.

Examples

Estimate format
- Scope: must-have vs nice-to-have
- Range: best / likely / worst
- Confidence: high / medium / low
- Risks: top 3 unknowns with mitigation

Tradeoffs

• Speed vs certainty: move fast, but make the unknowns visible.
• Local vs global optimization: align estimates to shared dependencies.
• Stability vs iteration: update estimates when new facts emerge.

Signals I look for

• Teams surface risks before stakeholders ask.
• Estimates get tighter after spikes and de-risking work.
• Roadmaps change based on new information, not wishful thinking.

Anti-patterns

• Point estimates treated as commitments.
• Hidden scope and untracked dependencies.
• Confidence levels that never change.