18.1 ROM, Summary-Level and Detailed Estimates: Preparing and Reviewing
Key Takeaways
- Blueprint tasks 1.E, 1.C, and 1.D ask a cost professional to prepare ROM and summary-level estimates but to review detailed estimates, matching the division of labour on most owner teams.
- Estimate class is set by scope definition maturity alone; effort and elapsed time do not upgrade a Class 5 estimate, and quoting one with Class 2 accuracy language is a serious misrepresentation.
- A ROM must be delivered as a range with the method, source, date, and exclusions stated — a bare point value invites the organization to treat a screening figure as a commitment.
- Detailed estimate review runs top-down in seven steps beginning with completeness against scope documents and basis-of-estimate review, not with checking arithmetic.
- Roughly 20% of line items carry about 80% of estimate value, so review the top decile in depth and sample the tail; reconciliation to the previous estimate is the technique that forces every movement to be explained.
18.1 ROM, Summary-Level and Detailed Estimates: Preparing and Reviewing
Three consecutive blueprint tasks describe the estimating products a CCP is expected to produce and to check: 1.E prepare rough order of magnitude (ROM) estimates, 1.C develop summary-level cost estimates, and 1.D review detailed project estimates. Note the verbs. The blueprint asks a cost professional to prepare the fast, high-level products and to review the detailed ones — which is the actual division of labour on most owner teams, where detailed estimating sits with a specialist estimating group or with a contractor.
1. Matching the Product to the Maturity
Chapter 2 covered the AACE 18R-97 classification system. This section is about using it as a decision tool: given where the project is, what product should exist, and what should it be trusted for?
| Product | Typical class | Scope definition | Usual method | Legitimate use |
|---|---|---|---|---|
| ROM | Class 5 | 0–2% | Capacity factor, analogous, order-of-magnitude ratio | Screening; go/no-go on a portfolio; sizing a study budget |
| Summary-level | Class 4–3 | 1–15% / 10–40% | Factored equipment, parametric models, unit rates against preliminary quantities | Budget authorization, funding request, feasibility |
| Detailed / definitive | Class 2–1 | 30–70% / 50–100% | Bottom-up quantity takeoff with priced resources | Bid, contract, control baseline |
[!IMPORTANT] The class is set by scope definition, not by effort or deadline. A team that spends six weeks on a project with 3% scope definition has produced a very carefully constructed Class 5 estimate. Effort does not upgrade class; only scope maturity does. Reporting a Class 5 estimate with Class 2 accuracy language is the single most damaging misrepresentation an estimator can make.
2. Preparing a ROM Estimate
A ROM's purpose is speed with an honest range. Discipline comes from three habits:
- State the method, the source, and the date. "Capacity-factored from Plant X (2023 actual, $/bbl-day basis), exponent 0.65, escalated on index 240 to 288." A ROM without a stated basis cannot be defended or improved.
- Quote a range, never a point. A Class 5 estimate is a range. Delivering "$42,000,000" invites the organization to treat it as a commitment; delivering "$34M to $59M, most likely $42M" communicates what the number actually is.
- List the exclusions explicitly. Owner's costs, land, permitting, escalation beyond a stated date, currency exposure, and interest during construction are the usual omissions. If they are excluded, say so in the same breath as the number.
Worked ROM
A client wants a screening figure for a 40 MW gas-fired peaking plant. The database holds a 25 MW unit completed for $27,500,000 two years ago; the index has moved 310 to 341; the capacity exponent for this plant type is 0.72.
| Step | Calculation | Result |
|---|---|---|
| Capacity scale | 40 / 25 = 1.60; 1.60^0.72 = 1.394 | factor 1.394 |
| Scale the cost | $27,500,000 x 1.394 | $38,335,000 |
| Escalate | x (341 / 310) = x 1.10 | $42,169,000 |
| Report as Class 5 | −15% / +40% around the point | $35.8M to $59.0M |
The deliverable is the range and the basis, with the point value identified as the most likely outcome rather than as the answer.
3. Summary-Level Estimates
A summary-level estimate is built from real but preliminary quantities, factored equipment costs, or parametric models. Three things make it defensible:
- A quantified basis. Even at Class 4, major equipment counts, gross areas, and bulk quantities should be listed rather than implied.
- Explicit factors. If bulks are factored at 1.8 times equipment cost, that factor is written down with its source, not buried in a spreadsheet cell.
- Contingency derived from the class and the risk register, not a habitual round percentage. Chapter 14 covers the derivation; the point here is that "we always use 15%" is not a basis.
4. Reviewing a Detailed Estimate (Task 1.D)
Reviewing someone else's detailed estimate is a structured search for a small number of recurring failures. Work top-down; do not start by checking arithmetic.
| Review step | The question | What it catches |
|---|---|---|
| 1. Completeness against the scope documents | Is every scope item in the estimate, and is every estimate item in the scope? | Missing systems; scope creep priced in silently |
| 2. Basis of estimate review | Are the assumptions, exclusions, and sources stated and reasonable? | Undocumented assumptions that later become claims |
| 3. Benchmark and ratio check | Do the totals and internal proportions match the peer set? | Structurally wrong allocations (see 17.1) |
| 4. Quantity check on the big items | Do the top items by value reconcile to drawings or models? | Takeoff errors where they matter |
| 5. Pricing check | Are unit rates current, and consistent with the market and location? | Stale rates; wrong location factor |
| 6. Indirects, escalation, contingency | Are they present, separately identified, and derived? | Missing indirects; contingency double-counted |
| 7. Reconciliation to the previous estimate | Why has the number moved since the last class? | Unexplained drift |
Pareto discipline
On a typical capital estimate, roughly 20% of the line items carry about 80% of the value. Reviewing every line at equal depth wastes the review budget on items that cannot move the total. Sort by value, review the top decile in detail, and sample the tail.
[!TIP] Reconciliation is the most powerful review technique. A bridge from the previous estimate to the current one — scope changes, quantity changes, rate changes, escalation, and contingency changes, each with a value — forces every movement to be explained. Any residual "unexplained" line in that bridge is exactly where the estimate's real problem lives.
A project has 3% scope definition. The estimating team spends six weeks producing a highly detailed spreadsheet build-up and presents the result as a definitive estimate suitable for contract award. What is the correct assessment?
A cost engineer is asked to produce a screening estimate for a 40 MW peaking plant. The historical database holds a 25 MW unit completed two years ago for $27,500,000, the relevant cost index has moved from 310 to 341, and the capacity exponent for this plant type is 0.72. What should the deliverable be?
A cost professional has three days to review a contractor's 4,000-line detailed estimate before an award recommendation. Which review approach is most defensible?