Cost Estimation

Numbers before money is committed: estimate from elements, state the class and range, and track every estimate against its actual — that's how estimates get good.

When to use

• Predicting cost of work, equipment, or purchases before committing.
• Reviewing vendor quotes (should-cost).
• Building rates and reference data from history.

Method

1. Class the estimate: concept/ROM (wide range, ±30–50% reality), budget (±15–20%), definitive (±5–10%) — label it, so nobody treats ROM as a quote.
2. Method per situation: bottom-up (elements) when design exists; parametric (unit rates, cost curves) when history exists; analogous (adjust a comparable explicitly) for one-offs.
3. Elements, not lump sums: material, labor (hours × rates incl. burden), equipment, subcontract, freight, indirects — each line traceable.
4. Uncertainty with three points: optimistic/likely/pessimistic for shaky items → expected value + spread; ranges always, precision theater never.
5. Escalation & currency: inflation/fx/lead-time impacts stated; quote validity dates captured.
6. Should-cost: for big items, build an independent benchmark and compare against vendor quotes; reconcile gaps (scope? rates? margin?).
7. Quote discipline: review exclusions and assumptions line by line; a quote without exclusions is a trap; validity period respected.
8. Feedback loop: track estimate vs actual per item; update rates; publish variance lessons (that's the compounding skill).

Quality gates

• Basis document: class, method, assumptions, exclusions, estimator.
• Ranges (not points) for material uncertainty; escalation handled.
• Quotes' exclusions reviewed; divergences explained.
• Estimate-vs-actual tracked and fed back into rates.

Red lines

• No single-number precision on uncertain scope; no vendor quote accepted without scope review; no padding hidden inside rates.

Related

project-budgets · budgeting · engineering-work-packages (scope stability).