Construction Scope of Work Checklist

There are two ways to review a scope of work. One is to read it and ask whether it seems complete. The other is to run it against a list.

The first method is how almost everyone does it, and it fails predictably. A reader assessing a document holistically finds the problems he is primed to find, reads past the sections he is comfortable with, and mistakes fluency for completeness. A twenty-page exhibit with polished boilerplate and no interface matrix reads better than a nine-page exhibit that covers everything, and holistic review will pass the wrong one.

This checklist is the second method. It is built to be run against a drafted exhibit before it goes out, section by section, with a specific question at each point. Eight parts, roughly two hundred checks, organized so you can run the whole thing on a major package in about ninety minutes and the critical subset in twenty.

One important distinction before you start. This checklist inspects your own document, not the design documents. Checking the drawings and specifications for unassigned or ambiguous scope is a separate exercise with a separate list. What follows assumes that work is done and asks a different question: does the exhibit you drafted actually say what you meant, bind what you intended, and hold up when somebody argues with it?

How to Use This Checklist

Run it against a draft, not a final. The point is to find problems while editing is cheap. An exhibit reviewed after issue can only be corrected by addendum, and after execution only by amendment.

Every check gets a disposition, not a checkmark. Four permitted answers: compliant, revised, not applicable with a stated reason, or accepted risk with a named approver. A tick tells you somebody looked. It does not tell you what they decided, and six months later that distinction is the whole point.

Use a reviewer who did not draft it. This is the single most important procedural rule here. A drafter reads what he meant; a fresh reader reads what is written. In my experience a second reader finds roughly twice as many defects, and the ones he finds are the consequential ones, because the drafter’s blind spots cluster exactly where his assumptions were strongest.

Run Parts E through G on the set, not the document. Parts A through D inspect one exhibit. Parts E, F, and G inspect the whole coordinated set of exhibits for a project, which is where a different and nastier class of failure lives.

Table 1. Checklist Parts and When to Run Them

Key Definitions

Exhibit. The scope of work attachment to a subcontract, commonly Exhibit B. The document this checklist inspects.

Component. One of the sixteen structural parts a complete scope of work contains. Part B checks their presence.

Matched pair. An inclusion in one exhibit with its corresponding exclusion in each affected neighboring exhibit. Both halves must exist.

Double assignment. The same obligation appearing as an inclusion in two or more exhibits without being decomposed as an interface. You will pay twice, or neither party will perform it.

Universal exclusion. An obligation excluded in every exhibit that touches it. It belongs to nobody.

Verb set. The specific performance verbs attached to an obligation: furnish, install, connect, coordinate, verify, test, commission. Precision here prevents the most common category of scope argument.

Disposition. The recorded outcome of a check. Compliant, revised, not applicable with reason, or accepted risk with an approver.

Objectives

• Confirm the exhibit contains every structural component, not just the legally familiar ones
• Catch drafting defects that create arguments regardless of how good the underlying scope decisions were
• Verify the exhibit set is collectively coherent, which no single-document review can do
• Prevent boundary loss during negotiation, which is where most scope control is quietly surrendered
• Keep the executed scope accurate as the project changes
• Produce a record of what was reviewed and decided, usable in a dispute

Part A: Pre-Drafting Readiness

Do not open the template until these are true. Drafting from an incomplete basis produces an exhibit that has to be rewritten rather than edited.

Table 2. Pre-Drafting Readiness Checks

Part B: Component Completeness

Sixteen components. Present, consciously addressed, or explicitly marked not applicable. Nothing silent.

Table 3. Component Completeness Checks

Score the exhibit. Most available templates score around seven to nine of sixteen, with absence clusters at B.7, B.8, B.10, B.12, and B.14. If your template follows the same structure, unfortunately, you have the average template. This is not a compliment.

Part C: Language Quality

This part ignores what the exhibit says and inspects how it says it. These are drafting defects, and they produce arguments even when every underlying scope decision was correct.

Table 4. Language Quality Checks

Part D: Trade-Specific Content

Component completeness confirms the structure. This part confirms the exhibit says the trade-specific things that trade’s exhibit must say. Run the relevant block per package.

Table 5. Trade-Specific Content Checks, Structure and Envelope

Table 6. Trade-Specific Content Checks, MEP and Specialties

If an item in the relevant block is not addressed anywhere in the exhibit, that is a finding. Silence is not exclusion, and an unaddressed item will be read one way by you and the other way by the subcontractor.

Part E: Cross-Package Reconciliation

This part inspects the set. It catches failures that are structurally invisible from inside any single exhibit, and it is the part almost everyone skips because it is genuinely hard to do across twenty-four separate Word documents.

Table 7. Cross-Package Reconciliation Checks

Part F: Pre-Issue

Run per package immediately before the bid package goes out.

Table 8. Pre-Issue Checks

Check F.8 deserves emphasis. Where you have assigned work based on how it is done in your market rather than on anything in the contract documents, that assignment has to be written into the exhibit. Your convention is not in the design documents, and a subcontractor from another market has no way to know it.

Part G: Pre-Execution

Run per package after negotiation, before signature. This is the last gate, and it is where scope control is most often surrendered.

Table 9. Pre-Execution Checks

Part H: Post-Execution Control

An executed exhibit begins going stale immediately. This part keeps it accurate.

Table 10. Post-Execution Control Checks

Check H.9 is what makes the checklist improve. A dispute is a diagnostic. Either way, the problem is either the check is nonexistent or it was skipped. In both cases, there are correctable issues. Only the case where it was skipped is about discipline.

Importance: What These Checks Are Preventing

Rework on commercial building work generally runs in the range of four to six percent of contract value, and a meaningful share of it traces to scope and coordination failures rather than defective workmanship. On a sixty million dollar project, one point of contract value is six hundred thousand dollars. You do not have to accept any specific published figure to see that a ninety-minute review of a document that governs a twelve million dollar mechanical subcontract is a favorable trade.

The behavioral effect matters as much as the arithmetic. Ambiguous exhibits do not distribute risk neutrally. Subcontractors who read the narrow interpretation carefully win the package and recover the difference through change orders. Bidders who price honestly lose. Every project bought out on vague exhibits filters your subcontractor pool one package at a time toward the adversarial end, and that filtering is slow, invisible, and hard to reverse.

Stakeholders

Table 11. Who Runs Which Part

The superintendent contributing to Part D is deliberate. Trade-specific content questions about hoisting, sequencing, access, and interfaces get answered incorrectly when the office answers them alone.

Technology Integration

Table 12. Checklist Execution by Method

Parts G and H barely move, and that is worth saying plainly. Pre-execution is a commercial conversation and post-execution control is a discipline problem. No tool performs either.

Part E is the one that changes character entirely. Comparing inclusion and exclusion lists across twenty-four documents is combinatorial, which is why it gets skipped, which is why double assignments and universal exclusions reach bidders routinely. When exhibits are generated from one structured register, the same check becomes a query and takes half an hour.

AI-Assisted Opportunities

The checks that resist manual execution all share a property: they require comparing content across an entire document set rather than reading one document carefully. That is a volume problem.

Preconstruction AI platforms address it upstream. iFieldSmart AI, for instance, extracts every note from every sheet with its exact drawing location, assigns trade responsibility and CSI division per obligation, and holds the result as a queryable register from which trade-specific scope documents are generated in editable Word. Because every exhibit is a filtered view of one dataset rather than an independently authored document, Part E becomes a set of queries, Part B components are present by construction, and Part A.8 becomes verifiable rather than asserted.

Table 13. Checklist Parts Mapped to AI Capability

Implementation

• Run Part B against your current template today. Score it out of sixteen. Most offices land at seven to nine, with absences at B.7, B.8, B.10, B.12, and B.14. That score is your baseline.
• Add Part C.1 immediately. Search every issued exhibit for “provide.” It costs twenty minutes per document and it addresses the most common drafting defect in the industry.
• Build the missing components into the template rather than into the checklist. A component the template prompts for is a component that gets drafted. A component only a checklist asks about is a component that gets skipped in week three.
• Make Part E a gate before issue, with a named reviewer who did not draft. Two to four hours per project.
• Make Part G a gate before signature, with written authority for the project manager to reject clarifications.
• Customize Part D to your market. Trade boundaries vary by region and office. Go through Tables 5 and 6 with your two most experienced estimators and mark which items your packages normally carry. That editing session converts a generic reference into your own scope convention.
• Instrument Part C and Part E as automated checks when your document tooling allows it, because those two are the parts most vulnerable to compliance decay.
• Run Part H.10 on your next two closeouts, retroactively if necessary. Mapping past scope disputes to components tells you which parts matter most in your specific market.

Table 14. Implementation Roadmap

Best Practices

Table 15. Checklist Discipline Best Practices

Common Mistakes

Table 16. Common Failures in Scope of Work Review

Applied Examples

Class A office core and shell, 340,000 square feet. Part A.14 drove everything. The work letter contradicted the drawings on ceiling grid extent and base building sprinkler head count. Because the readiness check required obtaining and analyzing the work letter before drafting, the conflict was found while it was still a boundary decision. Three affected exhibits carried explicit limits and matched exclusions. Found later, that is a three-party argument during tenant fit-out.

Regional hospital expansion, 190 beds. Part B.14 and B.15. Mechanical seismic restraint engineering was assigned to the mechanical subcontractor, the deferred submittal was assumed to be part of it, and special inspection was assigned to nobody. Special inspection on that scope is a condition of occupancy in the jurisdiction. Caught at component review it was a line item. Caught at inspection it is a schedule event with liquidated damages attached.

Hyperscale data center, 60 MW. Part E across twenty-eight packages found fourteen double assignments and six universal exclusions. Two universal exclusions were anchorage engineering for owner-furnished equipment, excluded by every mechanical and electrical bidder because none of them bought the equipment. That is the classic orphan on equipment-dense work, and E.2 is the only check that finds it.

Food grade manufacturing plant. Part B.10 and Part H.3. Process vendor documents arrived seven weeks after buyout by design. The exhibits carried named allowances for equipment support, hygienic panel interfaces, and process piping attachment, each with stated exclusions and a reconciliation method. Allowance balance reporting at each pay application meant the eventual reconciliation was a draw rather than a claim.

University laboratory building, 11 floors. Part E.4. Interface matrix identity across five packages touching casework and fume hood connections. On the prior project the same interfaces had been authored separately in each exhibit and had drifted into five different versions, which turned a coordination question into a five-party dispute. Differencing the matrix across all five exhibits took ten minutes.

Multifamily podium, 240 units. Part D on rough carpentry. The exhibit required a blocking schedule by room type with defined counts, because framers price only what is detailed and blocking is the recurring quantity-variable item on repetitive residential work. Seventy-one blocking obligations became a quantity rather than a risk, and the priced number came in below the allowance the team had carried.

Highway interchange, design-bid-build. Part D on earthwork and Part E.5. The exhibit addressed the trenching, bedding, and backfill boundary explicitly, and flow-down completeness caught that erosion control maintenance duration had been excluded by both the earthwork and site utility packages, leaving a permit condition with no owner.

Historic warehouse to hotel, 140 keys. Part C.7 and Part B.10. “As required” appeared 26 times across the draft exhibits, and “match existing” appeared 34 times in the source documents with no standard, extent, or acceptance criteria. The language checks forced each into either a defined quantity, a unit price, or a stated acceptance standard. On adaptive reuse this is the difference between hundreds of discoveries being arithmetic and being negotiations.

Frequently Asked Questions

How long does the full checklist take on a major package?

About ninety minutes for Parts A through D and F on a large mechanical or electrical exhibit, plus two to four hours for Part E across the whole set once per project, plus thirty minutes for Part G per package at execution. The critical subset, meaning B.7, B.8, B.10, B.12, B.14, C.1, and all of Part E, runs in about twenty minutes per exhibit.

Which checks should we adopt first if we can only do some?

C.1, B.7, and all of Part E. Search for “provide,” confirm the interface matrix exists and is decomposed, and reconcile the set for double assignments and universal exclusions. Those three address the drafting defect that generates the most arguments, the missing component that causes the most field disputes, and the failure class that no single-document review can find.

How is this different from a scope gap analysis checklist?

Different target document. A gap analysis checklist inspects the design documents, asking whether any obligation in the drawings and specifications is unassigned, double-assigned, or ambiguous. This checklist inspects the exhibit you drafted, asking whether it says what you meant and binds what you intended. The first is diagnosis of the source documents. The second is quality control on your own output. You need both, and doing one well does not substitute for the other.

Should subcontractors ever see this checklist?

Not the checklist, but share what it produces. The interface matrix, deliverables matrix, delegated design table, and allowance and unit price schedules should all go to bidders, including the rows showing other parties’ responsibilities. Keep the reconciliation findings and the component scoring internal, since those reflect your own quality control.

Is this appropriate for small projects?

Yes, scaled. On a project under ten million dollars, run Part A.6 through A.9, all of Part B, C.1 through C.10, the relevant Part D block for your three or four major packages, and Parts E.1 through E.3. That is perhaps three hours total. Small projects carry less margin to absorb surprises, which makes the review more valuable per dollar, not less.

How do we adapt this for renovation work?

Emphasize three checks. A.1 expands to include existing conditions documentation and a demolition limits matrix. C.7 becomes one of the highest-frequency findings, since “as required” and “match existing” proliferate on renovation documents. And B.10 becomes the central component rather than a supporting one, because unforeseen conditions dominate and unit prices are the mechanism that keeps discoveries from becoming negotiations.

What if the exhibit scores poorly and there is no time to fix it?

Disposition the findings as accepted risk with a named approver, and quantify the exposure. An exhibit issued with five known component gaps and a written record of who accepted them is a managed risk. The same exhibit issued with nobody having looked is an unmanaged one, and the difference matters both commercially and internally.

Who should be the independent reviewer?

Whoever did not draft, with enough construction knowledge to judge Part D. In smaller offices this is usually a peer preconstruction manager or the contracts lead. In larger ones a designated reviewer role works better, because the same person doing it across many projects develops pattern recognition the individual drafters cannot.

How do we prevent the checklist from becoming a ritual?

Two mechanisms. Build the checks into the template and into your tooling so the structure performs the check rather than a person remembering to. And run H.10 retrospectives so items that never produce findings come off the list and new patterns go on. A list nobody has revised in three years has stopped being a tool.

Does Part E work if exhibits are authored as separate Word documents?

It works, badly, and it is the honest reason most teams skip it. Comparing inclusion and exclusion lists across twenty-four independently authored documents is combinatorial. It can be done with a spreadsheet extract, and on projects up to about twelve packages that is manageable. Beyond that, generating exhibits from a single structured register is the practical answer.

What is the most commonly missed check across all projects?

B.12, coordination and site obligations, followed closely by E.2, universal exclusions. B.12 is missed because it reads as boilerplate and it is not, and everything left unstated defaults to the contractor. E.2 is missed because bid leveling structurally cannot find it: if every bidder excluded the same item, leveling reports perfect agreement.

Can we run these checks against a contract lifecycle management system?

Partly. Those platforms easily address version control, clause libraries, and approval routing, which supports Parts B, F, and H. They normally don’t have structured data linked to source documents, which is what Parts A.8, D, and E require. The two functionalities are quite different and complement each other.

Expert Recommendations

Score your current template against Part B this week. Twenty minutes, out of sixteen, and the score will tell you more about your scope exposure than any amount of discussion will.

Search every exhibit you have issued this year for “provide.” You will find dozens of instances, most of them on install obligations, and each one is a live argument waiting for a subcontractor who reads carefully.

Build the missing components into the template rather than relying on the checklist to catch them. Structure survives schedule pressure. Memory does not.

Name an independent reviewer role and give it authority to hold an exhibit from issue. A reviewer who can only advise will be overruled the first time bids are due Thursday.

Give project managers written authority to reject subcontractor clarifications. The pressure at award is commercial and personal, and people need documented cover to push back.

Customize Part D with your two most experienced estimators before your next bid. An hour of their time converts a generic reference list into your organization’s actual scope convention, and it is the difference between findings people trust and findings people dismiss.

Log every scope dispute against the check that would have caught it. One field, and it is the only mechanism that turns a static list into one that gets better every year.

Using This From Here

A checklist is an unglamorous instrument. It does not require software, headcount, or a change initiative. It requires deciding in advance what you will inspect, and then inspecting it the same way every time, including on the Thursday when bids are due.

What makes it effective on scope of work documents specifically is that the defects are not random. They cluster in the same five missing components, the same dozen language habits, and the same two reconciliation failures, across projects and across markets. That regularity is exactly what a list is for.

Score your template against Part B, run C.1 on five issued exhibits, and make Part E a gate before your next bid package goes out. Within three projects, the list will fit your office better than anything you could have purchased, because the items on it will be the ones that actually hurt you.