Cost Estimate Classification: Understanding Class 1–5 Estimates in Capital Projects

A cost estimate is only as useful as the decision it supports. A number produced from a capacity factor at concept stage has no business anchoring a final investment decision, and a detailed bottom-up estimate is overkill for a screening study. The discipline that keeps these uses separate — and keeps stakeholders honest about the uncertainty baked into any number — is cost estimate classification.

The AACE International framework defines five classes that tie an estimate’s maturity to the project definition behind it, the methodology used, and the accuracy range it can reasonably claim. This article walks through the five classes, how accuracy narrows as engineering develops, how classes map to decision gates, and the recurring mistakes that turn a useful estimate into a misleading one.

Why Estimate Classification Matters

Every cost estimate carries two numbers: the expected value and the range of uncertainty around it. Classification exists because the second number is routinely ignored. A Class 5 estimate produced at 1% engineering carries an accuracy range of roughly -50% to +100%. Quoted without that range and locked into an executive presentation, it can commit an organisation to a project whose realistic outcome is anywhere between half and double the sanction figure.

Classification forces three things into the conversation: how much of the project is actually defined, which methodology produced the number, and what accuracy band belongs to it. With those three stated, stakeholders can size contingency, judge decision risk, and set expectations about how far the number is likely to move as the project develops. The AACE framework is the most widely adopted system because it is methodology-based rather than tied to a specific industry phase gate, spanning project definition from less than 2% to more than 65%.

The Five Estimate Classes

The five classes form a continuum of maturity, each serving a different decision and using different methods. The defining attributes sit in the table below, which preserves the AACE accuracy bands and project-definition ranges.

Class 5, the order-of-magnitude estimate, relies on cost-capacity factors, cost-per-square-metre rates, and benchmark data from analogue projects; it answers whether an idea is worth further study. Class 4 supports feasibility analysis using preliminary layouts, early equipment selection, and approximate quantities. Class 3 is the sanction estimate, typically produced during Front-End Engineering Design (FEED) and used for the final investment decision, with credibility resting on structured methodology and a documented basis-of-estimate. Class 2 is the control estimate, built from detailed quantity takeoffs once engineering is 30–70% complete, and becomes the baseline against which execution is tracked. Class 1 is the definitive estimate, reserved for the most mature stage when drawings, quantities, procurement pricing, and contractor quotes are fully in hand.

How Accuracy Narrows Through the Lifecycle

The accuracy band is not a function of the estimator’s skill; it is a function of what is known about the project at the time. A capital project moves through a predictable sequence — concept, feasibility, FEED, detailed design, construction — and at each stage more information becomes available. Equipment lists firm up, plot plans are frozen, procurement packages go out for pricing, contractor bids come back. Each step removes uncertainty from a specific category of cost.

The asymmetry of the early-stage bands is worth noting. Class 5’s -50% to +100% is deliberately skewed upward because at concept stage the risks that compound cost — scope growth, site-specific complications, missed quantities, escalation — all push in the same direction. As definition improves, the band narrows and symmetrises, reaching -10% to +15% at Class 1 because the remaining uncertainty is execution-driven rather than definition-driven.

This is also why early-stage estimates require higher contingency. A Class 5 estimate with a 15% contingency is pretending to be a Class 3 estimate. Contingency should be sized to the residual uncertainty described by the class, not to the comfort level of the person presenting the number.

Mapping Classes to Decision Gates

Classification is the link between estimating and governance. Each class supports a specific capital-project decision, and the gates in a stage-gate model are designed around the accuracy required at each stage. Project screening runs on Class 5, feasibility approval on Class 4, sanction on Class 3, execution authorization typically on Class 2, and construction control against Class 1.

The failure mode is using the wrong class for the wrong gate. Approving a $1B project on a Class 5 estimate commits capital to a range of $500M to $2B — the accuracy band says so explicitly. Teams that force a sanction decision on Class 4 numbers absorb the difference through contingency draw, scope cuts, or overruns that surface in execution. Mature organisations enforce a minimum class at each gate and require the basis-of-estimate to document how it was achieved.

Common Classification Mistakes

Four mistakes recur across capital projects, and each one hollows out the value of the framework. The first is treating early estimates as budgets: a Class 5 or Class 4 number quoted without its accuracy band gets locked into executive expectations, and every downstream estimate is then pressure-tested against the original concept figure rather than against current project definition.

The second is stripping the accuracy range from communication. Stakeholders see a single deterministic number, not the -30% to +50% band that belongs to it, and contingency discussions collapse into negotiation rather than analysis. The third is mixing methodologies without structure — detailed takeoffs for civils, parametric factors for mechanicals, benchmark rates for electricals — producing a number that looks Class 3 but behaves like Class 4. The class of a blended estimate is governed by its weakest component, not its strongest.

The fourth is undersizing contingency for the class. Each class carries an implied minimum contingency that reflects the uncertainty in its accuracy band. Setting a 10% contingency on a Class 4 estimate does not make it a better estimate — it makes it a Class 4 estimate likely to overrun by the difference between the band and the contingency.

Key Takeaways

• Estimate classification ties maturity, methodology, and accuracy together — an estimate without its class and accuracy range is an incomplete number.
• The AACE Class 1–5 framework spans 0–2% to 65–100% project definition, with accuracy bands narrowing from -50%/+100% to -10%/+15%.
• Each class is matched to a specific decision gate; using the wrong class for a gate exposes the project to the full width of the accuracy band.
• Contingency must be sized to the class, not to the audience; under-contingent early-stage estimates are the most common source of sanction-era overruns.
• The class of a blended estimate is governed by its weakest component — methodology discipline matters more than the accuracy of any one line.

Related reading

• Parametric Estimating in Early-Stage Capital Projects
• Cost Contingency in Capital Project Estimates