How AEC Engineers Prepare a Defensible Opinion of Probable Construction Cost

What Is an Opinion of Probable Construction Cost?

You've seen the term on every public works project you've touched— and if a bid has ever come in significantly higher than your estimate, you've also felt its weight. An opinion of probable construction cost (OPCC) is a licensed engineer's professional judgment about what a project will likely cost to construct, prepared at a specific point in the design process. It's called an "opinion"— a deliberate contractual choice— because the engineer has no control over bid market conditions, material prices, or competitive pressures at the time of bidding.

Under the EJCDC E-500— the industry-standard engineering services contract— OPCCs represent the engineer's "best judgment as an experienced and qualified professional generally familiar with the construction industry"¹. The engineer isn't committing to a number. They're providing the best available cost information given what exists on paper at that point in design.

OPCCs are typically requested at the following stages of a project²:

• Inception / Feasibility: go/no-go decision support before significant design investment
• 30% design: budget authorization for capital programming
• 60% design: confirmation that the project remains within budget as design details emerge
• 90% design: final pre-bid cost check for owner approval and contractor bid comparison

That progression matters. Each OPCC reflects a different level of design completeness— which is exactly why accuracy expectations change dramatically across the design lifecycle. And so does understanding what the word "opinion" legally protects, where the AACE Class 1–5 framework calibrates those expectations, and which AI tools now fit inside this workflow without replacing the professional judgment at its center. This article covers all three.

Why Engineers Say "Opinion"— The Legal and Professional Significance

The word "opinion" is a deliberate professional and contractual choice. Under the EJCDC E-500, the engineer "cannot and does not guarantee that proposals, bids, or actual Construction Cost will not vary from opinions of probable Construction Cost"¹. That language does specific work: it signals where the engineer's professional responsibility ends and market forces begin.

Practitioners draw this line carefully. From the Eng-Tips engineering community⁴:

"An estimate implies a higher level of accuracy than an opinion of cost... an opinion of cost is usually qualified heavily and may vary 25%."

The National Society of Professional Engineers (NSPE) reinforces this position. NSPE formally opposes requirements that engineers guarantee their services, because such requirements shift "an inequitable or unacceptable amount of responsibility to the professional engineer"⁵. Their standard disclaimer language: "The Engineer does not guarantee the accuracy of such estimates as the Engineer has no control over the cost of labor and material, competitive bidding, or market or other conditions."⁵

And here's the nuance practitioners often miss. The "opinion" label is not a blanket liability waiver— standard of care still applies. Some public agencies require root cause analysis when a bid varies from the engineer's estimate by more than 10%⁶. An OPCC that was wildly unreasonable for the design stage— or that omitted obvious cost drivers— remains professionally vulnerable regardless of how it's titled. Documentation of assumptions is what makes that threshold defensible. If owner and engineer need greater certainty, the EJCDC also provides for engaging an independent cost estimator as a formal contract add.

The AACE Cost Estimate Classification System: Understanding Class 1 Through Class 5

Every cost estimate has an accuracy ceiling. AACE International— the American Association of Cost Engineers— publishes the industry's standard for where those ceilings are. The system classifies estimates into five classes, with Class 5 representing the earliest, least-defined stage and Class 1 representing near-complete design. Per AACE Recommended Practice No. 17R-97³: "the maturity level of definition is the sole determining (primary) characteristic of class."

That's an important frame. Accuracy is a function of design completeness, not engineer skill. A Class 3 OPCC from a rigorous firm will carry the same accuracy range as one from a less experienced firm— because the bounds are set by how complete the design is, not by who built the spreadsheet.

Per AACE Recommended Practice No. 17R-97, at 80% confidence.¹

One practical note for civil AEC work: the AACE classes map most cleanly to EPC and industrial projects. For civil projects— roads, water mains, municipal buildings— the milestone column above is the more useful anchor. Accuracy within each class also varies by project type; water/wastewater projects tend to show lower Class 3 variance than transportation projects, but the AACE framework provides the accepted baseline. Early-phase estimates at Class 5 can be up to an order of magnitude off the final bid amount⁶— which is why owners who treat a feasibility-phase OPCC as a budget commitment are setting themselves up for a painful conversation.

How Engineers Build an OPCC at Each Design Phase

Engineers use three primary methods to prepare an OPCC, and the appropriate method depends on how complete the design is at the time. Early phases use parametric estimating— cost per unit of capacity— later phases use unit-price assembly and detailed quantity takeoffs as drawings become specific enough to support them.

For most US firms, that database is RSMeans. Published by Gordian, the 2025 edition contains over 23,000 lines of costing data updated annually, with City Cost Indexes and square-foot costs for 160 common commercial building configurations⁷. Engineers apply City Cost Indexes to adjust national averages to local markets— a step that matters, since the spread between a mid-sized Midwestern city and a coastal urban market can run 30–40% on the same scope of work⁷.

Smaller civil firms relying on their own historical bid data often find local conditions reflected more accurately than national averages. Unit prices from similar completed projects in the same market, adjusted for current escalation, carry calibration a national database can only approximate. That's local expertise doing its work.

Contingency and Escalation: Pricing In What You Don't Know

Contingency is a structured allowance for cost uncertainty— it is not a slush fund, and it is not optional. Engineers size contingency to reflect how much design definition remains: the less complete the design, the higher the contingency.

Per AACE guidance⁸, contingency allowances by design phase run roughly:

• Study/Feasibility phase: 20% or more of base costs
• ~30% design: approximately 15%
• Final design: approximately 10%
• After bids received: approximately 5%

Escalation is a separate calculation. But it compounds the contingency issue. Construction costs change between the date an OPCC is prepared and the date a contractor actually bids the project. When there's a meaningful time gap between OPCC delivery and expected bidding, engineers carry an escalation factor. The US construction cost escalation forecast for 2025 was approximately 2.5%— down significantly from a 9% peak in early 2023— but ongoing tariff risks on steel and concrete remain a material variable⁹¹⁰.

In practice: a 30% design OPCC for a project bidding in 18 months might carry 15% contingency and 3.5% escalation applied to base costs. Both should be documented separately from the base estimate numbers, not buried inside them. The owner needs to understand what's a known cost and what's a planned allowance for uncertainty. Documentation discipline— separating base costs, contingency, and escalation— is also what makes AI-assisted estimates reviewable.

Where AI Tools Help Engineers Prepare a Faster, More Defensible OPCC

Building those estimates— quantity takeoffs, unit-price lookups, market calibration— historically took days of careful work. AI tools are now compressing two specific parts of that process without compromising the professional judgment at its center.

Quantity takeoff automation. Togal.AI uses computer vision to identify and measure building elements directly from PDF plan sets— wall lengths, floor areas, component counts. The vendor claims 98% accuracy on floor plans¹²; independent industry analysis puts the range at 80–98% on well-drawn commercial construction sets, with 70–90% reduction in quantity takeoff time¹³. Best results come from complete, well-drawn commercial building sets at 60%+ design. For civil and infrastructure work— utilities, roadway plan sets, grading plans— accuracy at the same confidence level hasn't been established; run a project-specific pilot before committing to Togal.AI for civil OPCC support.

Conceptual AI estimating. Gordian Flash AI Estimating, launched in early 2026, uses contextual AI trained on RSMeans data to generate high-level conceptual estimates from uploaded construction documents— specs, plans, or scope narratives¹¹. Gordian describes it as capable of producing "defensible, high-level estimates in minutes, not hours or days." For Class 4–5 work where detailed drawings don't exist, this compresses days of unit-price research into a first-pass number the engineer can review and refine.

Peer-reviewed research supports the direction. Artificial neural network and multiple regression methodologies "provide sizeable improvements over current levels of prediction accuracy for construction costs"¹⁴. AI estimating adoption among contractors stood at approximately 19% in 2024, with BuildingRadar projecting near 38% by 2026; of early adopters surveyed, 68% reported savings above $50,000 annually¹³. One source, not independently verified by AGC, ENR, or Dodge Construction Network— but the direction is consistent with what we see in the field.

For AEC firms exploring where AI fits in their practice, AI implementation for engineering firms can help identify the right entry points before committing to a software subscription or workflow change.

What AI Cannot Replace in an Opinion of Probable Construction Cost

AI automates the mechanical steps of OPCC preparation— counting items, looking up unit prices, generating first-pass numbers. What it does not do is make the professional judgments that determine whether those numbers are defensible.

Most engineering firms evaluating AI cost tools are solving for speed. That's a reasonable goal. But the gap that actually shows up in disputed OPCCs is underdocumented assumptions— and no software fills that gap.

What AI cannot do in an OPCC context:

• Document assumptions and exclusions that define the scope boundary— what is and is not in the cost
• Apply professional judgment on site conditions: geotechnical risk, access constraints, utility conflicts
• Size contingency appropriately for project-specific risk beyond the phase-based guidelines
• Assess local market calibration: whether RSMeans reflects what will actually be bid in this specific project environment
• Navigate owner-engineer dynamics and scope creep that affect the budget reality

These five areas are where the engineer's professional value is highest— and where an AI tool positioned correctly behind the judgment layer becomes most productive.

The garbage-in problem is real. AI estimating tools are only as accurate as their training data. For rural markets, unusual project types, or materials with volatile pricing in the current tariff environment, RSMeans may not reflect what will land in actual bid tabs. And the engineer who seals and certifies the OPCC remains professionally responsible for its content regardless of which tools generated the underlying numbers⁶.

When evaluating AI tools for your practice, the useful question is which workflow steps benefit from automation and which require judgment that can't be delegated.

How to Document an OPCC to Protect Your Firm and Serve the Owner

A well-documented OPCC protects the engineer and serves the owner. The documentation should be specific enough that anyone reading it understands exactly what is and is not included in the cost— not because the contract requires it, but because it's the difference between a defensible professional opinion and an exposed number.

Some agencies require root cause analysis when a bid varies from the engineer's estimate by more than 10%⁶. That analysis is only possible if the OPCC was built to support it. Every OPCC should include:

1. Design phase reference: State explicitly which phase (e.g., "prepared at 30% design documents dated [date]")
2. AACE class designation or equivalent design maturity reference
3. Accuracy range: Explicitly state the expected range (e.g., "−10% to +30% at 80% confidence per AACE Class 3")
4. Assumptions list: What is included in the scope
5. Exclusions list: What is not— owner-furnished items, hazardous materials remediation, permitting fees, right-of-way acquisition
6. Cost basis: Which database was used (RSMeans year; City Cost Index applied)
7. Escalation: The date through which costs apply and any escalation allowance
8. Contingency: The contingency percentage and the basis for selecting it⁸
9. Standard disclaimer: NSPE-aligned language per your contract template⁵

Before deploying AI tools into your estimation workflow, establishing AI governance policies for your practice ensures your documentation standards hold— particularly around which AI-generated outputs require engineer review before entering the OPCC.

Frequently Asked Questions About Opinions of Probable Construction Cost

What does opinion of probable construction cost mean?

A licensed engineer's best professional judgment of what a project will likely cost to construct, based on the level of design complete at the time. It carries no contractual price obligation— that is the domain of a contractor's bid. Under the EJCDC E-500, an OPCC represents the engineer's "best judgment as an experienced and qualified professional generally familiar with the construction industry"¹.

How accurate is an opinion of probable construction cost?

Accuracy depends entirely on design completeness, per the AACE classification system³. At conceptual/study phase (Class 5): expect ±30–50%. At 30% design (Class 3): roughly ±10–30%. At 90–100% design (Class 1): ±3–15%. All ranges are at 80% confidence per AACE RP 17R-97.

What contingency is standard in an OPCC?

Roughly 20% or more at study phase, reducing to approximately 10% at final design and approximately 5% after bids are received, per AACE guidance⁸. Contingency should be documented separately from the base cost estimate— the percentage reflects how much uncertainty remains in the scope, and it should decrease as design matures and unknowns resolve.

Can AI replace an engineer's cost estimate?

No. AI automates quantity extraction and unit-price lookup, but it does not replace professional judgment on assumptions, exclusions, site conditions, escalation, or local market calibration. The engineer who certifies the OPCC remains professionally responsible for its content regardless of which tools assisted in preparation.

What databases do engineers use to prepare OPCCs?

RSMeans, published by Gordian, is the dominant US cost database— 23,000+ line items updated annually, with City Cost Indexes for local market adjustment⁷. Engineers also use their own historical bid data for projects where local conditions diverge from RSMeans national averages.

Conclusion

An opinion of probable construction cost is both a professional artifact and a professional commitment. The word "opinion" signals not modesty but precision: accuracy is bounded by design completeness, and those bounds are knowable— that's what the AACE Class 1–5 system exists to clarify.

The AACE framework gives engineers and owners a shared language for accuracy expectations. AI tools give engineers speed at the mechanical stages without compromising the judgment that makes the OPCC defensible. Both are valuable, and neither replaces the other. But the estimate is only as good as the assumptions behind it— and those still require a licensed engineer to document and own.

If your firm is evaluating AI tools for cost estimation— or trying to build defensible OPCC documentation practices— Dan Cumberland Labs helps AEC firms identify where AI genuinely fits and where professional judgment remains the irreplaceable piece.

References

1. Engineers Joint Contract Documents Committee, "E-500: Standard Form of Agreement Between Owner and Engineer for Professional Services" (2014)— https://ejcdc.org/product/e-500-standard-form-of-agreement-between-owner-engineer-for-professional-services-rd-edition/
2. GBRIONLINE.org, "Engineering Cost Estimate Phases— OPC Discussion" (Current)— https://www.gbrionline.org/?p=123759
3. AACE International, "Recommended Practice No. 17R-97: Cost Estimate Classification System" (1997, current)— https://epcland.com/aace-estimate-classification-system/
4. Eng-Tips Engineering Community, "Engineers' Estimates— Forum Discussion" (Ongoing)— https://eng-tips.com/threads/engineers-estimates.289476
5. National Society of Professional Engineers, "Professional Liability— Position Statement" (Current)— https://www.nspe.org/nspe-advocacy/explore-issues/professional-policies-and-position-statements/professional-liability
6. Zweig Group, "Liability for Cost Estimates/Cost Overruns" (Current)— https://zweiggroup.com/blogs/news/liability-for-cost-estimates-cost-overruns
7. Gordian/RSMeans, "2025 Building Construction Costs Book" (2025)— https://www.rsmeans.com/products/2025-building-construction-costs-book
8. AACE International / EPCland, "AACE Estimate Classification System— Contingency Application" (Current)— https://epcland.com/aace-estimate-classification-system/
9. Turner & Townsend, "US Construction Market Intelligence— Q2 2024 Escalation Forecast" (2024)— https://marketintelligence.turnerandtownsend.com/usmi-q2-2024-summer/escalation-forecast
10. Oxford Economics, "Cost Escalation Pressures Are Easing But Key Risks Remain: Construction and Infrastructure" (2024)— https://www.oxfordeconomics.com/resource/cost-escalation-pressures-are-easing-but-key-risks-remain-construction-and-infrastructure/
11. Gordian, "Flash AI Estimating— Product Page" (2026)— https://www.gordian.com/flash-ai-estimating/
12. Togal.AI, "How It Works" (Current)— https://togal.ai/how-it-works
13. BuildingRadar, "How Is AI Used in Construction Cost Estimating?" (2024)— https://www.buildingradar.com/construction-blog/how-is-ai-used-in-construction-cost-estimating
14. Technical Journal, UET Taxila, "The Influence of Artificial Intelligence on Practices in Construction Cost Estimation" (2024)— https://tj.uettaxila.edu.pk/index.php/technical-journal/article/view/1933