The most damaging thing that happens to a construction estimate is that a number produced with a stated ±40% range gets quoted three months later as the budget, without the range. Everything that follows — approvals, funding, scope commitments — is then built on a precision the estimate never claimed.
Estimate classification exists to prevent that. It ties a number to the maturity of the information behind it and to an honest accuracy range.
The AACE framework
The AACE International classification, widely used in North America and increasingly internationally, defines five classes by the degree of project definition.
Class 5 — Concept screening. Project definition around 0–2%. Prepared from capacity factors, parametric models, and analogy to completed projects. Typical accuracy range in the region of −30% to +50%, sometimes wider. Used to decide whether an idea is worth exploring.
Class 4 — Study or feasibility. Definition around 1–15%. Based on preliminary layouts, equipment lists, and elemental rates per unit area. Typical range around −20% to +30%. Used for feasibility and initial funding requests.
Class 3 — Budget authorization or control. Definition around 10–40%. Based on developed design, semi-detailed unit costs, and takeoff of major quantities. Typical range around −15% to +20%. This is normally the estimate that becomes the control budget.
Class 2 — Control or bid. Definition around 30–75%. Detailed takeoff from substantially complete documentation, with subcontractor pricing on major packages. Typical range around −10% to +15%.
Class 1 — Check estimate or definitive. Definition around 65–100%. Full detailed takeoff, firm quotations, and detailed construction planning. Typical range around −5% to +10%.
The percentages are indicative, not contractual. Actual ranges vary with project type, market volatility, and how much of the scope is genuinely defined. A Class 3 estimate for a repeat warehouse is far tighter than a Class 3 estimate for a hospital refurbishment.
Why the class must travel with the number
The single most useful practice in estimating is refusing to issue a number without its class, its range, its basis, and its date. An estimate should always be presented as a range with a stated central value, never as a single figure.
Concretely, every estimate should carry a basis of estimate document naming:
- The estimate class and expected accuracy range
- The documents used, with revision and date
- The pricing date and the escalation assumptions
- Included and excluded scope, stated explicitly
- Design allowances and contingency, separately identified
- Assumed procurement route and contract form
- Market conditions assumed
- Known risks not priced
The exclusions list is the part most often thin and most often the source of dispute. "Excludes abnormal ground conditions" is not an exclusion; it is a placeholder.
Contingency, allowance, and risk
Three distinct provisions are frequently conflated:
Design development allowance covers the growth in scope that reliably occurs as design matures. It is not risk; it is the known tendency of incomplete design to become more, not less. It reduces as design progresses and should approach zero at Class 1.
Contingency covers identified risks that may or may not occur, quantified where possible. It is properly derived from a risk register rather than applied as a flat percentage, though flat percentages remain common at early classes.
Management reserve covers scope changes and events outside the estimate's assumptions. It typically sits with the client, not in the construction estimate.
Presenting a single blended percentage hides which of these is being consumed, which is exactly the information a project needs during delivery.
Escalation
An estimate is priced at a date. Construction occurs later. Escalation from the pricing date to the midpoint of construction should be calculated explicitly, using published indices where available, and stated as a separate line rather than absorbed.
In volatile markets this is one of the largest single adjustments in the estimate, and one of the most frequently omitted. Where lead times are long and commodity exposure is significant, the escalation assumption deserves its own sensitivity analysis.
Elemental versus trade structure
Estimates are structured either elementally — by functional building element such as substructure, superstructure, envelope, services — or by trade package, matching how the work will be procured.
Elemental structure supports benchmarking and design decision-making: it answers "is our envelope cost reasonable for this building type." Trade structure supports procurement and cost control during delivery.
Mature estimating practice maintains a mapping between the two, so a number can be viewed either way. Committing to only one creates friction at every handover between design and delivery.
Benchmarking as a control
Every estimate should be tested against benchmarks before issue:
- Cost per square metre or square foot, by building type and market
- Elemental cost per unit area, compared against a benchmark set
- Key quantity ratios — concrete volume, steel tonnage, glazing area, and services cost as a proportion of the total
- Preliminaries and overhead as a proportion, checked against comparable projects
Benchmarks do not validate an estimate. They identify where it is unusual, which is where the review should concentrate. An element that is 40% away from benchmark is either wrong or reflects something specific that should be explainable in one sentence.
Communicating uncertainty
The presentation problem is real: clients want a number, and a range invites the reader to remember only the low end.
What works in practice is presenting the central value prominently with the range immediately alongside, stating explicitly what would move the number in each direction, and showing how the range is expected to narrow at the next design stage. Framing the range as a schedule of decreasing uncertainty is more persuasive than framing it as an admission of imprecision.
Reconciling between estimate stages
The most scrutinised document in cost management is not any single estimate but the reconciliation between two of them. When a Class 3 estimate is higher than the Class 4 that preceded it, someone will ask why, and "the design developed" is not an answer.
A proper reconciliation attributes every difference to a category:
Design development. Scope that was always intended and is now defined. This should consume the design development allowance rather than adding to the total.
Scope change. Work added or removed by decision. This is the client's to own and should be visible.
Quantity refinement. Same scope, better measurement.
Rate change. Same quantity, different price — market movement, better information, or corrected earlier error.
Escalation. Movement in the pricing date.
Error correction. Stated plainly. Concealing a corrected error inside "design development" destroys credibility permanently when it is found.
Presenting this as a bridge — starting total, each movement, ending total — converts a difficult conversation into a factual one. It also disciplines the estimating team, because each category has to be defensible.
Estimating in a volatile market
Periods of material and labour volatility change what a good estimate looks like. Adjustments that matter:
Shorten the validity period. An estimate valid for ninety days in a stable market may be valid for thirty in a volatile one, and it should say so.
Price commodity exposure separately. Identify the elements most exposed — steel, copper, aluminium, timber, fuel — and show their proportion of the total, so the client can see what is at risk.
Use sensitivity analysis rather than larger contingency. Showing the total at several material-price scenarios is more useful than a single inflated figure, because it tells the client what would have to happen for the number to move.
Date every rate. Rates carried from six months ago in a volatile market are a systematic error, not a minor one.
Watch lead times, not just prices. Extended lead times affect programme, and programme affects preliminaries, which are often a larger movement than the material price itself.
Frequently asked questions
How long does a detailed estimate take? For a mid-size commercial project at Class 2 or 1, several weeks including takeoff, pricing, subcontractor enquiry, and review. Compressing it forces reliance on rates rather than measured quantities, which widens the real range regardless of what the classification claims.
Should the estimate include profit and overhead? It should identify them explicitly rather than embedding them in unit rates, so the client can see the cost structure and the estimator can adjust them independently of the measured work.
What causes the largest estimating errors? Scope omission, not pricing error. Work that was never measured cannot be mispriced — it simply appears later as a change.
Related reading: Quantity Takeoff: Manual, Digital, and Model-Based Compared · 5D BIM: Linking Cost to the Model Without Losing Control · NRM, CESMM and POMI: Measurement Standards for UK and International Work
Vantage CAD Services provides quantity takeoff and estimating support at every design stage, with structured basis-of-estimate documentation. Contact info@vantagecadservices.com or +1 (512) 543-0831.