Most people in construction use “scope gap” to mean one thing: work that got left out. That definition is comfortable, and it is wrong, and being wrong about it is expensive.
Work almost never gets left out. Design teams are good at their jobs, and a coordinated commercial set describes the building. What goes wrong is assignment. The obligation exists in the documents, sometimes in three places, and no trade package clearly carries it. Or two packages carry it. Or one package furnishes it, and nobody installs it. According to the specs, the drawings don’t show it, and the estimator, who quantifies from drawings, never see it.
For four different reasons, four different things fail to happen, and four different options exist for how each thing is detected. Four different things depend on each other to some degree, but because most teams view each document in a set as a standalone instance, the reviewers expect to find certain things and are shocked to discover the other things in month five.
This article breaks scope gaps into ten distinct types. For each: what it actually is, why it occurs, where it conceals itself, what it expenses, how to entrap it. When you incorporate this taxonomy on how your team reviews documents, you will cease searching for the ambiguous term called “gaps” and will commence performing ten specific checks.
Why Classification Changes Outcomes
Consider two findings on the same project. On sheet A-101, a general note reads “provide access as required for concealed equipment.” On sheet M-201, a note reads “duct insulation per specification,” and the mechanical package includes insulation while the insulation package also includes it.
Both are scope gaps in casual usage. They have nothing in common operationally. The first is an ambiguous attribution problem, found by screening language, fixed by an RFI or a documented convention. The second is an overlap, found by querying for multi-assigned obligations, fixed by a scope narrative exclusion. Different search, different owner, different remedy.
A reviewer looking for “gaps” finds whichever type she is mentally primed for. A reviewer running ten typed checks finds all ten, because each check has a defined target.
Table 1. The Ten Gap Types at a Glance
| # | Gap Type | Core Failure | Primary Detection Method |
|---|---|---|---|
| 1 | Unassigned obligation | No package carries the work | Division coverage crosswalk |
| 2 | Overlap and double assignment | Two packages both carry it | Multi-assignment query |
| 3 | Ambiguous attribution | Responsible party described, not named | Language pattern screen |
| 4 | Interface split | Furnish, install, coordinate, or verify unassigned | Interface decomposition grid |
| 5 | Cross-discipline displacement | Obligation on one discipline’s sheets, another’s trade | Trace by assigned trade across sources |
| 6 | Specification-only obligation | Required in specs, absent from drawings | Bidirectional spec reconciliation |
| 7 | Drawing-only obligation | Shown on drawings, no governing specification | Bidirectional spec reconciliation |
| 8 | Delegated design fragmentation | Engineering, submittal, or inspection unassigned | Delegated design trigger screen |
| 9 | Boundary gap | Obligation falls across a contractual boundary | Boundary matrix per boundary type |
| 10 | Exclusion cascade | Every bidder excluded the same item | Exclusion reconciliation per package |
Two of these deserve a note up front. Type 5 is the highest-frequency gap on commercial building work and the one manual review handles worst. Type 9 is the highest-severity, because contractual boundaries involve parties outside your subcontract structure and therefore outside your control.
| KEY TAKEAWAY Ten types, ten checks, ten different remedies. Most preconstruction teams would be best served by no longer viewing scope gap review as a single activity. These teams should assign a type and target to each inspection of the scope gap review. |
Gap Type 1: The Unassigned Obligation
What it is. An obligation stated in the contract documents that no bid package carries. The genuine orphan.
Why it happens. Almost always because the obligation’s content spreads across many packages, so no single subcontractor recognizes it as theirs. Divisions 07, 09, 10, 12, and 31 generate these at a much higher rate than any others, and the reason is structural rather than technical. Firestopping appears in six divisions. Blocking appears wherever anything gets mounted. Nobody’s core trade identity includes them.
The second cause is package boundary drift. Your office has always put access panels in drywall. This project’s estimator, newer, assumed mechanical. Neither wrote it down. The panels end up in neither.
Where it hides. General notes on architectural sheets. Division 01. Division 10 specialties. Substrate preparation obligations. Anything described as “provide” without a system context.
What it costs. Variable, but the distribution is bad. Small items get absorbed. Large ones surface after installation of adjacent work, which means the fix carries removal, replacement, and delay on top of the original scope.
How to catch it. A division-to-package crosswalk. Every division present in the project manual is a row, every bid package is a column, and each cell is marked full, partial, or none. Any division with no full or partial mark anywhere is unexamined. Any division marked partial in every column has an unassigned remainder that somebody needs to name.
| FIELD REALITY The unassigned rate on an exhaustively extracted commercial set typically runs one to three percent of obligations. If your log shows zero unassigned items, the analysis is not clean, it is incomplete. A healthy review finds problems. |
Gap Type 2: Overlap and Double Assignment
What it is. Two or more packages both carry the same obligation.
Why it happens. Broad spec section language, defensive subcontractor scoping, and the natural tendency of a GC to make sure something is covered by putting it in two places. That last one feels prudent and is not.
What it costs. Two ways, and the second is worse than the first. You paid twice, which is a straightforward margin loss. And when both subs believe the other one has it, neither does it, which is the same outcome as Type 1 with the added irritation of having paid for it. Overlaps look like the safe failure mode. They are not.
Where it hides. Insulation between mechanical and the insulation trade. Firestopping between the penetrating trade and the firestop specialist. Painting of exposed conduit and ductwork. Ceiling access provisions. Excavation and backfill between earthwork and the utility trades. Cleaning between each trade’s final clean and the GC’s general cleaning obligation.
How to catch it. Query the obligation register for items assigned to more than one trade where no interface was defined. Then, separately, compare the inclusion lists across adjacent packages during buyout. The second check catches overlaps that entered through subcontractor proposals rather than through your assignment.
Table 2. Recurring Overlap Pairs on Commercial Work
| Obligation | Packages That Both Claim It | Resolution Convention |
|---|---|---|
| Duct and pipe insulation | Mechanical, plumbing, insulation specialist | Assign by system, state the split explicitly in both narratives |
| Penetration firestopping | Each penetrating trade, firestop specialist | Assign by penetration type, not by trade |
| Painting of exposed conduit, duct, hangers | Painting, each MEP trade | Assign to painting; exclude explicitly in MEP narratives |
| Trench excavation and backfill | Earthwork, each utility trade | Earthwork performs, utility trades supply and lay, joint hold point |
| Equipment pads | Concrete, mechanical, electrical | Concrete performs, equipment trade provides layout and templates |
| Access panels | Drywall, ceilings, mechanical, plumbing | Assign by assembly type, furnish and install stated separately |
| Final cleaning | Each trade, GC general conditions | Each trade cleans its own work; GC performs final clean; define the handoff |
| Fire alarm devices and interfaces | Electrical, fire alarm, mechanical | Assign by device and by interface point |
| Sleeves and penetration formwork | Concrete, each penetrating trade | Penetrating trade furnishes, concrete places, GC verifies before pour |
Gap Type 3: Ambiguous Attribution
What it is. The obligation names a responsibility but not a party. “By others.” “Coordinate with adjacent trades.” “As required.” “By installing contractor.”
Why it happens. Design documents are written to describe a completed building, not to divide a subcontract structure the designer has never seen. “By others” is a perfectly reasonable thing for an architect to write. It becomes a gap only when a GC’s buyout treats it as an assignment.
What it costs. This is the type that most reliably produces change orders, because ambiguity does not stay neutral in a competitive market. Subcontractors who read documents carefully price the narrow interpretation, win the package, and recover the difference later. Vague scope systematically selects for the bidders most willing to argue with you in month seven.
Where it hides. In language, not in subject matter, which is why it needs its own screening pass independent of what the obligation is about.
Table 3. Ambiguous Attribution Patterns and What They Actually Mean
| Phrase as Written | What the Bidder Reads | What You Need It to Say |
|---|---|---|
| “By others” | Not me | Named trade or GC-held line |
| “As required” | Whatever is drawn, and nothing more | Quantity, location, and standard |
| “Coordinate with adjacent trades” | Attend the meeting; install nothing | Who coordinates, who installs, who verifies |
| “Field verify” | Risk transferred; scope unchanged | Verification obligation with a named party and a deadline |
| “By installing contractor” | Whoever that turns out to be | The installer, named |
| “Provide for” / “allow for” | Leave room; supply nothing | Defined capacity, count, or dimension |
| “Rough-in only” | Stop at the wall | The precise stopping point and who continues |
| “Furnish only” | Deliver to the site | Who receives, stores, protects, and installs |
| “Match existing” | The cheapest thing that resembles it | Standard, extent, and acceptance criteria |
| “Typical” | This detail everywhere | Where it applies and where it does not |
How to catch it. Screen the full text of every note and specification paragraph for the patterns above, before assignment. It is the cheapest check available and the one that best predicts change orders. Log every hit even when it looks harmless.
| COMMON MISTAKE When teams face uncertain attribution, the clarification is requested from the design team. Often, the design team provides restated responses. Teams then accept the item as closed. Write RFIs that force a decision. “Confirm which trade furnishes and installs access panels at fire dampers in rated assemblies, and confirm panel type per 08 31 13” gets an answer. “Please clarify access panel requirements” gets you the note you already read. |
Gap Type 4: The Interface Split
What it is. An obligation involving two or more trades where the responsibilities have been named at the trade level but not decomposed. Both subs know they are involved. Neither has a complete understanding of the endeavor.
Each multi-trade obligation has four separate responsibilities designed to answer the following questions: who supplies the materials, who does the installation, who arranges the location and order, and who checks the correctness. Naming two trades answers none of those four questions.
Why it happens. Because “mechanical and drywall” feels like a resolution. It is recorded in the log, the log shows no open item, and the analysis appears complete. This is probably the most common false resolution in preconstruction.
What it costs. Field disputes, almost always at the worst possible moment. Interface obligations are usually sequence-critical, so the argument happens when one trade is ready to proceed and the other is not.
Where it hides. Everywhere two systems meet. Equipment connections. Penetrations. Supports and attachments. Substrate handoffs. Anything owner-furnished.
Table 4. Interface Decomposition Examples
| Obligation | Furnish | Install | Coordinate | Verify |
|---|---|---|---|---|
| Access panel at fire damper | Drywall or mechanical, by assembly | Drywall typically | Mechanical provides locations | Superintendent, then AHJ |
| Housekeeping pad at pump | Concrete | Concrete | Plumbing provides layout and anchor template | GC before pour |
| Sleeve through rated wall | Penetrating trade | Penetrating trade | GC | Firestop specialist and AHJ |
| Final connection to owner-furnished equipment | Owner | Named trade per system | GC | Commissioning agent |
| Ceiling grid at sprinkler head | Ceilings and fire protection | Each own component | VDC or ceilings | Superintendent |
| Roof curb for RTU | Mechanical or steel, by type | Roofing installs flashing | GC | Roofing warranty inspector |
| Low-voltage backbox and pathway | Electrical | Electrical | Low-voltage trade provides locations | Commissioning |
| Seismic restraint at duct | Delegated engineer and fabricator | Mechanical | Structural review | Special inspector |
How to catch it. Build the grid. Any multi-trade obligation with a blank cell is an open finding, regardless of how many trades are named. Fill it in with a superintendent in the room, because the answers depend on how work actually gets sequenced.
Gap Type 5: Cross-Discipline Displacement
What it is. An obligation appears on one discipline’s drawings but belongs to a different discipline’s trade. Electrical work described on mechanical sheets. Concrete work described on plumbing sheets. Carpentry work described on architectural sheets.
Why it happens. Designers annotate where the reader needs the information, which is correct practice. A mechanical engineer noting “provide power for VAV boxes” on M-301 is being helpful to the mechanical contractor. The problem is entirely on the buyout side: estimators read by discipline. The electrical estimator reads E sheets. Nobody reads M sheets looking for electrical obligations.
What it costs. This is the highest-frequency gap type on commercial building work by a wide margin. A mid-size commercial mechanical set will routinely carry twenty to forty obligations belonging in the electrical package.
Where it hides. Predictably. The pairs repeat on every project.
Table 5. Cross-Discipline Displacement, Recurring Pairs
| Obligations Appear On | Belong To | Typical Items | Frequency |
|---|---|---|---|
| Mechanical drawings | Electrical | VAV power, thermostat and sensor wiring, disconnects, control panel feeds | Very high |
| Architectural drawings | Carpentry | Blocking at accessories, mounts, handrails, casework, grab bars | Very high |
| Architectural drawings | Electrical, low voltage | Device and plate locations, floor boxes, AV backboxes, pathway | Very high |
| Architectural and mechanical drawings | Firestopping, sealants | Penetrations and joints in rated assemblies, head-of-wall | Very high |
| Mechanical and plumbing drawings | Concrete | Housekeeping pads, equipment bases, pits, trench drains | High |
| Structural drawings | MEP trades | Sleeves, embeds, penetration limits, hanger attachment | High |
| Roof plans | Steel, roofing, mechanical, electrical | Curbs, supports, anchors, walkway pads, tie-off points, lightning protection | High |
| Electrical drawings | Drywall, ceilings | Openings, backing, access, grid coordination | Moderate |
| Civil drawings | Electrical, plumbing | Site lighting circuits, utility transitions at the five-foot line | Moderate |
| Vertical transportation drawings | Electrical, structural, drywall | Hoistway prep, sill angles, machine room power and cooling | Moderate |
How to catch it. Assemble scope by assigned trade rather than by drawing discipline, then query each pair above and review the results as a batch. The batch matters. Ten displaced items reviewed together make the pattern obvious in a way no single item does.
| EXPERT TIP If you have one hour and a searchable scope register, run pairs one through four in this table. On commercial building work those four pairs account for the majority of recurring cross-discipline findings, and they are all fast to query. |
Gap Type 6: The Specification-Only Obligation
What it is. A requirement stated in the specifications with no representation on the drawings.
Why it happens. Specifications describe performance, standards, and quality. Drawings describe geometry and location. Plenty of legitimate obligations have no geometry to draw. Hangers and supports, grounding and bonding, seismic restraint, insulation thickness, testing and balancing, commissioning participation, and special inspections all live primarily in text.
The failure is on the estimating side. Estimators quantify from drawings, because that is where quantities are. An obligation with no drawing representation gets no quantity, and something with no quantity gets no dollars.
What it costs. Consistently significant, because these are exactly the obligations that carry hidden engineering, testing, and inspection costs. Seismic restraint on a mechanical package can run six figures on an equipment-dense project and appear nowhere on a plan.
Where it hides. Specification part 1 and part 3, mostly. Submittal requirements, quality assurance, field quality control, and commissioning sections.
Table 6. High-Frequency Specification-Only Obligations
| Spec Area | Obligation | Why It Gets Missed | Typical Exposure |
|---|---|---|---|
| 23 05 29 | Hangers, supports, seismic bracing | No plan representation; delegated engineering attached | High |
| 26 05 26 | Grounding and bonding beyond code minimum | Assumed included in general electrical | Moderate |
| 22 07 00 / 23 07 00 | Piping and duct insulation extent | Partial plan representation understates quantity | Moderate to high |
| 07 84 00 | Penetration firestopping | Schedule without locations; no quantity basis | High |
| 01 91 00 | Commissioning participation by each trade | Nobody’s core scope; effort underestimated | Moderate |
| 01 45 00 | Special inspections and testing | Split between GC, owner, and trades | Moderate |
| 09 22 16 | Non-structural framing deflection criteria | Triggers delegated engineering | Moderate to high |
| 23 05 93 | Testing, adjusting, and balancing | Assumed part of mechanical; often a separate agency | Moderate |
| 01 77 00 | Closeout, O&M, training, attic stock | Deferred until it is a retention issue | Low to moderate |
| 05 05 23 | Welding qualification and inspection | Inspection cost and schedule unassigned | Moderate |
Note: *Spec Area references follow the CSI MasterFormat specification numbering system. *01 45 00 – Quality Control, 01 77 00 – Closeout Procedures, 01 91 00 – Commissioning, 05 05 23 – Welding Connections, 07 84 00 – Firestopping, 09 22 16 – Non-Structural Metal Framing, 22 07 00 – Plumbing Insulation, 23 05 29 – Hangers and Supports for HVAC Piping and Equipment, 23 05 93 – Testing, Adjusting, and Balancing (TAB) for HVAC, 23 07 00 – HVAC Insulation, and 26 05 26 – Grounding and Bonding for Electrical Systems.
How to catch it. Bidirectional reconciliation. Walk the specifications and ask, for each substantive requirement, whether the drawings show it. The answer is frequently no, and frequently that is fine, but each one needs a conscious decision about whether it was priced.
Gap Type 7: The Drawing-Only Obligation
What it is. Something shown on the drawings with no governing specification section.
Why it happens. Late design additions, addenda that add drawing content without adding spec sections, and details borrowed from a previous project.
What it costs. Usually less than Type 6, but in a more irritating form. The item gets installed to whatever standard the installing subcontractor chooses. Then the submittal gets rejected, or the installation gets rejected, and the argument is about whether the sub had any basis to know better. He did not.
Where it hides. Specialty items on architectural details. Site features on civil and landscape drawings. Anything added by addendum.
How to catch it. The other direction of the same reconciliation pass. For each item shown on the drawings, confirm a spec section governs it. Where none does, either get one by RFI or state the standard in the subcontract scope narrative yourself.
Gap Type 8: Delegated Design Fragmentation
What it is. A single delegated design requirement creates three separate obligations, and they get assigned to three different parties or to none. The three are: the engineering itself, the deferred submittal, and the special inspection.
Why it happens. Because the requirement reads as one thing. “Provide framing designed by a licensed engineer” sounds like one obligation to assign to the framer. It is three, and the framer prices the first, assumes the second is paperwork, and has no idea the third exists.
What it costs. The engineering is usually priced. The submittal is usually absorbed. The inspection is the one that surprises people, and it surprises them late, because special inspection is frequently a condition of occupancy. A missed special inspection discovered near substantial completion is a schedule event, not a cost event, and schedule events on commercial projects carry liquidated damages.
Where it hides. Anywhere the specification says engineer, design-build, performance, deferred submittal, or licensed professional.
Table 7. Delegated Design Triggers and Their Three Obligations
| System | Engineering | Deferred Submittal | Special Inspection |
|---|---|---|---|
| Cold-formed metal framing over height limit | Framer’s engineer | Framer | Frequently unassigned |
| Mechanical hangers and seismic restraint | Mechanical’s engineer | Mechanical | Frequently unassigned |
| Fire sprinkler hydraulic design | Fire protection contractor | Fire protection | AHJ plus special inspector |
| Curtain wall and storefront | Manufacturer’s engineer | Glazing | Structural observation |
| Precast connections | Precast engineer | Precast | Welding and bolting inspection |
| Steel connections where delegated | Fabricator’s engineer | Steel | Welding inspection |
| Handrails and guardrails | Fabricator’s engineer | Metals | Load testing where required |
| Canopies and trellises | Fabricator’s engineer | Metals | Structural observation |
| Suspension of ceilings in seismic zones | Ceiling manufacturer or engineer | Ceilings | Special inspection is required |
| Anchorage of equipment | Frequently unassigned | Frequently unassigned | Frequently unassigned |
The last row is serious. Owner-furnished equipment anchorage is one of the most reliably unassigned obligations in commercial construction, because the owner furnishes the equipment, the equipment vendor does not engineer the anchorage, and no trade package includes engineering for equipment it did not buy.
How to catch it. Screen the specifications for delegated design triggers, then build a three-column grid per trigger. Any blank cell is a finding.
Gap Type 9: The Boundary Gap
What it is. An obligation that falls across a contractual boundary rather than across a trade boundary. The parties involved are not all your subcontractors, which means you cannot resolve it by reassignment.
Why it happens. Boundaries are described in documents that are not the specifications and often not the drawings: work letters, tenant criteria manuals, utility agreements, owner-furnished equipment schedules, property line easements, and phasing agreements. These documents get treated as background rather than as primary sources.
What it costs. More than any other type, because the counterparty is not under your subcontract. A gap between two of your subs is a conversation. A gap between your work and a utility company’s work, or a tenant’s contractor’s work, is a negotiation with someone who has no obligation to help you.
Table 8. Boundary Types and Where the Gaps Concentrate
| Boundary | Governing Document | Recurring Gap Items | Severity |
|---|---|---|---|
| Base building to tenant improvement | Work letter, tenant criteria | Ceiling grid extent, sprinkler head count, HVAC distribution, panel capacity, restroom count | Very high |
| Contractor to utility company | Utility agreement, service application | Service transitions, transformer pads, meter locations, primary versus secondary, trenching | Very high |
| Owner-furnished to contractor-installed | OFCI schedule | Receiving, storage, protection, rigging, anchorage, final connection, startup, warranty | Very high |
| Property line and right of way | Civil drawings, easements, permits | Sidewalk, curb cut, utility transitions, restoration, traffic control | High |
| Phase to phase | Phasing plan, milestone schedule | Temporary conditions, partial systems, protection, temporary certificate of occupancy scope | High |
| Existing building to new construction | Renovation drawings, existing conditions | Tie-in points, capacity of existing systems, selective demolition limits | High |
| Process vendor to construction | Vendor scope documents | Foundations, supports, utilities to equipment, controls integration, validation | High |
| Design team to contractor, delegated | Specifications | Engineering, submittal, inspection, as noted in Type 8 | Moderate to high |
| Landlord to tenant, retail | Lease exhibit, work letter | Storefront, signage, HVAC tie-in, grease exhaust, utility metering | High |
Where it hides. In the documents nobody treats as contract documents. Get the work letter. Get the OFCI schedule. Get the utility agreement. Read them as primary sources and extract obligations from them the same way you extract from the specifications.
| WARNING On core and shell projects, the base-building-to-tenant-improvement boundary is the most consequential scope line in the project, and the work letter frequently contradicts the drawings. When they conflict, the conflict does not resolve itself. It resolves during tenant fit-out, when a GC is arguing with a tenant’s contractor and the developer is arguing with both. |
Gap Type 10: The Exclusion Cascade
What it is. Every bidder on a package excludes the same item. Bid leveling shows perfect agreement. The item belongs to nobody.
Why it happens. Subcontractors in a given market share a common reading of ambiguous documents, and they exclude defensively. Three electrical bidders who all exclude equipment pads are not conspiring. They are all applying the same regional convention, which happens to differ from what your documents assumed.
What it costs. The cost is ordinary. The detection failure is what makes this type dangerous, because the standard tool for finding scope problems during bid evaluation is leveling, and leveling is blind to this by design. Leveling compares bidders to each other. If they all agree, it reports agreement.
Where it hides. In the exclusion sections of subcontractor proposals, which get read for pricing implications and not for coverage implications.
How to catch it. Build an exclusion reconciliation sheet per package. List every exclusion from every bidder. Mark which ones appear on all proposals. Every universal exclusion is either assigned to another package or GC-held with dollars. There is no third option and no exclusion gets left unclaimed.
Table 9. Frequently Cascaded Exclusions by Package
| Package | Commonly Excluded by All Bidders | Where It Actually Belongs |
|---|---|---|
| Electrical | Equipment pads, cutting and patching, painting of raceway, temporary power beyond a stated duration | Concrete, drywall, painting, GC general conditions |
| Mechanical | Access panels, housekeeping pads, roof curbs, test and balance agency | Drywall, concrete, steel or roofing, separate agency |
| Plumbing | Housekeeping pads, trap primers, cutting and patching, final connection to owner equipment | Concrete, plumbing by clarification, drywall, negotiated |
| Drywall | Access panel furnishing, backing and blocking, substrate preparation | Mechanical or specialty, carpentry, preceding trade |
| Framing and carpentry | Blocking not detailed on drawings, plywood backboards | Requires a blocking schedule; electrical or low voltage |
| Roofing | Curbs and supports, walkway pads, tie-off anchors, lightning protection | Mechanical or steel, roofing by clarification, specialty |
| Concrete | Embeds furnished by others, sleeve furnishing, equipment templates | Steel or MEP trades, penetrating trade, equipment trade |
| Earthwork | Utility trenching, dewatering beyond a stated volume, import or export | Utility trades or earthwork by clarification, GC allowance |
| Fire protection | Attachment to structure beyond standard, escutcheons at special ceilings, fire pump power | Structural or by clarification, ceilings, electrical |
Relative Frequency and Cost
Not all ten types deserve equal attention. Allocate effort by expected value.
Table 10. Gap Type Frequency, Cost, and Detection Difficulty
| Gap Type | Relative Frequency | Cost per Occurrence | Detection Difficulty | Priority |
|---|---|---|---|---|
| 5, Cross-discipline displacement | Very high | Low to moderate each, high in aggregate | Low with a register, very high manually | 1 |
| 3, Ambiguous attribution | Very high | Moderate | Very low, it is a text search | 2 |
| 4, Interface split | High | Moderate to high | Low, but requires field judgment | 3 |
| 9, Boundary gap | Moderate | Very high | Moderate, needs the right documents | 4 |
| 6, Specification-only obligation | High | Moderate to high | Moderate | 5 |
| 8, Delegated design fragmentation | Moderate | High when inspection is missed | Low, trigger words are searchable | 6 |
| 1, Unassigned obligation | Moderate | Variable, worst tail | Low with a division crosswalk | 7 |
| 2, Overlap and double assignment | High | Low to moderate | Low with a register | 8 |
| 10, Exclusion cascade | Moderate | Moderate | Low, one sheet per package | 9 |
| 7, Drawing-only obligation | Low to moderate | Low | Moderate | 10 |
The ordering rewards types where detection is cheap relative to consequence. Types 3, 5, and 8 are all essentially searches. If your obligations are in a searchable register, they cost almost nothing to run and they cover the three highest-frequency categories.
Stakeholders and Detection Ownership
Table 11. Who Detects Which Gap Type
| Gap Type | Best Positioned to Detect | Why |
|---|---|---|
| 1, Unassigned | Preconstruction manager | Owns the division-to-package crosswalk |
| 2, Overlap | Estimator | Sees inclusion lists across adjacent packages |
| 3, Ambiguous attribution | Preconstruction manager or analyst | Text screen, no trade knowledge required |
| 4, Interface split | Superintendent | Knows how work sequences and who shows up |
| 5, Cross-discipline displacement | Preconstruction with VDC support | Requires querying across disciplines |
| 6, Specification-only | Estimator with spec discipline | Quantifies; feels the absence of a quantity |
| 7, Drawing-only | Project engineer during submittals | Discovers it when the standard is questioned |
| 8, Delegated design | Project manager or contracts | Owns submittal and inspection tracking |
| 9, Boundary | Project executive or precon director | Only role with visibility into the contract structure |
| 10, Exclusion cascade | Estimator | Reads every proposal |
Type 9 belongs at the executive level for a specific reason. Boundary gaps involve documents and counterparties that project-level staff often do not see, and resolving them frequently requires a commercial conversation rather than a technical one.
Required Documentation for Typed Detection
Table 12. Documents Required per Gap Type
| Document | Types It Detects | Consequence of Not Having It |
|---|---|---|
| Full drawing set with revision index | 1, 2, 4, 5, 7 | Analysis against superseded sheets |
| Complete project manual | 1, 6, 8 | Types 6 and 8 undetectable |
| Division 01 General Requirements | 1 | GC-held obligations unassigned |
| Addenda and bulletins | All | Register describes a set that no longer exists |
| Bid package definitions | 1, 2, 10 | No assignment destination; coverage uncomputable |
| Work letter or tenant criteria | 9 | Highest-severity boundary unexamined |
| OFCI equipment schedule | 4, 9 | Split-responsibility obligations invisible |
| Utility agreements and service applications | 9 | Utility boundary gaps surface during construction |
| Subcontractor proposals with exclusions | 10 | Exclusion cascade undetectable |
| Phasing plan | 9 | Phase-boundary obligations unassigned |
| Prior project gap log | All | No pattern library; each project starts from zero |
Note: Type numbers refer to the 10 Scope Gap Types used throughout this article: 1 – Unassigned Obligation, 2 – Overlap and Double Assignment, 3 – Ambiguous Attribution, 4 – Interface Split, 5 – Cross-Discipline Displacement, 6 – Specification-Only Obligation, 7 – Drawing-Only Obligation, 8 – Delegated Design Fragmentation, 9 – Boundary Gap, and 10 – Exclusion Cascade. All indicates the document supports detection across all ten scope gap types.
Technology and Detection
Table 13. Detection Method by Technology Approach
| Gap Type | Manual Review | Searchable Document Set | Structured Scope Register | AI Extraction and Query |
|---|---|---|---|---|
| 1, Unassigned | Partial, depends on reviewer | Partial | Good, crosswalk computable | Complete, coverage reportable |
| 2, Overlap | Poor | Poor | Good | Complete |
| 3, Ambiguous attribution | Partial | Good | Good | Complete across notes and specs |
| 4, Interface split | Good, judgment-driven | Good | Good | Candidates pre-identified |
| 5, Cross-discipline | Poor, reviewers read by discipline | Partial | Good | Complete, queryable by trade |
| 6, Specification-only | Partial | Partial | Good | Complete, bidirectional |
| 7, Drawing-only | Partial | Poor | Good | Complete, bidirectional |
| 8, Delegated design | Partial | Good, trigger words | Good | Complete |
| 9, Boundary | Depends on document access | Same | Same | Same, if boundary documents are ingested |
| 10, Exclusion cascade | Good | Good | Good | Good |
Two things are worth noticing in that table. Types 4 and 10 are handled well manually, because they are judgment and comparison tasks at manageable volume. Types 5, 6, and 7 are handled poorly by every manual method, because they require systematic comparison across a whole document set. That is the honest division of labor.
AI-Assisted Detection
The gap types that resist manual detection all share a property: they require examining the entire document set rather than the parts a reviewer thinks are relevant. That is a volume problem, and volume problems are what automation solves.
Agentic AI platforms built for preconstruction address this by extracting every note from every sheet with its exact drawing location, proposing a responsible trade and CSI division per obligation, and exposing the result as a queryable register. iFieldSmart AI works this way, including the ability to click an item in the register and land on the annotation that produced it, which is what makes an assignment defensible six months later when someone asks why.
Table 14. AI Use Case Matrix by Gap Type
| Gap Type | AI Query or Capability | Human Decision Required |
|---|---|---|
| 1, Unassigned | Report obligations with no trade assignment, by division | Assign, or GC-hold with dollars |
| 2, Overlap | Report obligations with multiple trade assignments and no defined interface | Adjudicate the split |
| 3, Ambiguous attribution | Pattern match every ambiguous phrase across notes and specs | RFI, reassignment, or allowance |
| 4, Interface split | Flag multi-trade obligations as decomposition candidates | Four responsibility answers |
| 5, Cross-discipline | Retrieve obligations by assigned trade across all source disciplines | Confirm package membership |
| 6, Specification-only | Compare spec requirements against drawing representation | Judge materiality and exposure |
| 7, Drawing-only | Identify drawn items with no governing spec section | Get a section or state the standard |
| 8, Delegated design | Detect triggers across the project manual | Assign engineering, submittal, inspection |
| 9, Boundary | Extract obligations from work letters and OFCI schedules alongside specs | Commercial negotiation |
| 10, Exclusion cascade | Compare exclusion sets across proposals per package | Reassign or GC-hold |
The conversational layer matters more than it sounds. Asking for electrical obligations appearing on mechanical drawings, or obligations where both plumbing and fire protection hold responsibility, or Division 23 items with no trade assigned, and getting a sourced answer that exports in one click, is what turns Types 1, 2, and 5 from projects into questions.
| IMPORTANT Automation changes where the human effort goes, not how much of it there is. The reviewer stops hunting and starts adjudicating, which is harder work requiring more experience. Teams expecting to need fewer people are usually disappointed. Teams expecting to cover the whole set usually are not. |
Implementation
- Adopt the taxonomy as vocabulary first. Have your team classify the last twenty scope change orders on a completed project by type. This takes an afternoon and it will immediately show which two or three types dominate in your market.
- Build the three cheap searches. Types 3, 5, and 8 are all text or query operations. Set them up as saved searches or standing queries so they run without anyone remembering.
- Make the interface grid a gate. No package awards with a blank cell on a multi-trade obligation. Type 4 is the one that most reliably prevents field disputes, and it is a policy decision rather than a tooling decision.
- Get the boundary documents at intake. Work letter, OFCI schedule, utility agreements, phasing plan. Type 9 is the highest-severity category and it is undetectable without them.
- Add the exclusion reconciliation sheet to bid evaluation. One sheet per package, an hour of work, and it closes the one gap type that leveling structurally cannot find.
- Classify field-discovered gaps at closeout. Every gap that reached the field gets a type. Aggregate over three projects and you will know exactly which check to strengthen.
Best Practices
Table 15. Best Practices by Gap Type
| Practice | Types Addressed |
|---|---|
| Build a division-to-package crosswalk on every project | 1 |
| Assemble scope by assigned trade, never by drawing discipline | 5 |
| Run the language screen before assignment, not after | 3 |
| Decompose every multi-trade obligation into four responsibilities | 2, 4 |
| Reconcile specifications and drawings in both directions | 6, 7 |
| Screen the project manual for delegated design triggers and build the three-column grid | 8 |
| Treat work letters, OFCI schedules, and utility agreements as primary documents | 9 |
| Build an exclusion reconciliation sheet for every package | 10 |
| State conventional assignments explicitly in the scope narrative | 1, 3 |
| Classify every field-discovered gap at closeout | All |
Note: Numbers correspond with 10 types of Scope Gap described in this article: 1. Unassigned obligation; 2. Overlap and double assignment; 3. Ambiguous attribution; 4. Interface split; 5. Cross-discipline displacement; 6. Specification-only obligation; 7. Drawing-only obligation; 8. Design fragmentation; 9. Boundary gap; and 10. Exclusion cascade. All indicates the practice helps address every scope gap type.
Common Mistakes
Table 16. Misdiagnoses and Their Cost
| Mistake | What Actually Happens | Correction |
|---|---|---|
| Treating all gaps as Type 1 | Nine detection methods never run | Adopt the taxonomy as working vocabulary |
| Assuming overlaps are harmless | Both subs assume the other has it; you paid twice for nothing | Treat Type 2 with the same seriousness as Type 1 |
| Resolving Type 3 with a vague RFI | Answer restates the ambiguity | Write RFIs that force a decision and name the parties |
| Recording two trades as an interface resolution | Type 4 remains open while the log shows it closed | Require all four responsibility answers |
| Reading the set by discipline | Type 5, the highest-frequency category, largely missed | Query by assigned trade across all sources |
| Quantifying only from drawings | Type 6 obligations receive no dollars | Walk the specifications independently |
| Assigning delegated engineering and stopping | Special inspection surprises you near occupancy | Three columns per trigger, all filled |
| Treating the work letter as background | Highest-severity boundary unexamined | Extract obligations from it like a spec section |
| Relying on bid leveling to find gaps | Type 10 invisible by construction | Exclusion reconciliation sheet per package |
| Not classifying field gaps at closeout | Same types recur indefinitely | One type code per field-discovered gap |
How the Types Show Up Across Sectors
Class A office core and shell, 340,000 square feet. Type 9 dominated. The work letter set ceiling grid at the core only, the drawings showed grid throughout the floor plate, and the sprinkler head count in the work letter assumed an open plan. Three conflicts, all on the base-building-to-TI boundary, all caught because the team extracted obligations from the work letter as a primary document. Discovered later, each becomes a three-party argument.
Acute care hospital, 240 beds. Types 8 and 9 together. Medical gas: the specification required a certified installer, the drawings showed rough-in on plumbing sheets, and the equipment schedule listed outlets as owner-furnished. That is a boundary gap and a verification gap in the same system, and on healthcare work the consequence is licensing rather than cost. Resolution separate parties for piping, outlets, alarms, certification, and third-party verification.
60 MW hyperscale data center, types 5 and 6. Types 5 and 6. Housekeeping pads appeared across M, E, and S sheets with three implied owners, which is Type 5 at a scale that matters given the equipment count. Separately, seismic restraint and hanger engineering appeared in the specifications with no plan representation, which is Type 6 with a Type 8 obligation buried in it. Both found in precon. Both would have been six-figure conversations in the field.
Advanced manufacturing plant, food grade. Type 9, process vendor boundary. Vendor documents arrived seven weeks after buyout, which was planned. The team flagged vendor-dependent obligations, carried a named allowance for equipment support, hygienic panel interfaces, and process piping attachment, and drew against it when the package landed. No claim, because the boundary was identified as a boundary rather than discovered as a surprise.
University science building, 11 floors. Type 5 on the architectural-to-carpentry pair produced 63 blocking obligations across five room types, against a framing package that had priced detailed blocking only. Converting findings into a room-type blocking schedule turned an unknown into a quantity.
Multifamily podium, 240 units. Types 5 and 2. Blocking again, plus a genuine overlap where both the framing and drywall packages carried backing at bath accessories. The overlap was found by querying multi-assigned obligations, which took minutes, and would have been paid twice.
Highway interchange, design-bid-build. Type 9, contractor-to-utility boundary. Conduit for future intelligent transportation systems appeared in the electrical special provisions and on the signal plans, with the trenching sitting between the electrical sub and the earthwork contractor. Resolved by writing trenching, bedding, and backfill into earthwork, conduit and pull boxes into electrical, and a joint hold point before backfill.
Historic warehouse to hotel, 140 keys. Types 3 and 9. “Match existing” appeared 34 times across the architectural set with no standard, extent, or acceptance criteria, which is Type 3 at volume. And the existing-to-new boundary needed a demolition limits matrix by area and system, without which selective demolition and trade-performed removals overlapped everywhere.
FAQ’s
Which gap type costs the most across a typical commercial portfolio?
Type 9, boundary gaps, has the highest cost per occurrence, because the counterparty sits outside your subcontract structure. Type 5, cross-discipline displacement, has the highest aggregate cost, because individual items are modest but the frequency is very high. If you are optimizing for total dollars, work Type 5. If you are optimizing against catastrophic single events, work Type 9.
Which type is most often mistaken for another?
Type 4 gets recorded as resolved when it is not. A log entry reading “mechanical and drywall” looks like an assignment and is actually an unresolved interface. This is probably the single most common false resolution in preconstruction, and it is invisible in any log that does not require four answers per multi-trade obligation.
Do these types apply outside commercial building work?
The taxonomy holds; the distribution shifts. On heavy civil, Type 9 dominates because of utility and right-of-way boundaries, while Types 1 and 2 are less common because bid item structures provide a coverage check. On industrial work, Type 9 at the vendor boundary dominates. On renovation, Type 3 spikes because of “match existing” and “field verify.” Same ten types, different weighting.
Can one obligation be more than one type?
Frequently, and it matters for detection rather than for classification. A specification requirement for seismic restraint with no plan representation and unassigned special inspection is Type 6 and Type 8 simultaneously. Classify by the check that would have found it earliest, and note the second type, because the remedy usually needs to address both.
What percentage of obligations end up in each type?
On an exhaustively extracted commercial set, rough proportions look like this: one to three percent unassigned, five to ten percent multi-trade or ambiguous, one to two percent specification-only with material exposure, and a handful of boundary and delegated design items per project. Absolute counts matter more than percentages. Forty to fifty items requiring a real decision on a mid-size project is normal.
How does this taxonomy relate to a scope gap analysis checklist?
The taxonomy explains why the checklist items exist. A checklist tells you to look for blocking on architectural drawings. The taxonomy tells you that is Type 5, that Type 5 is detected by querying across disciplines, and that the same method finds several other things. Use the taxonomy to either create or evaluate a checklist. Then use the checklist to perform the task at hand.
Which types can be detected before the bid documents are complete?
Types 8 and 9 largely can, because delegated design requirements and contractual boundaries are usually established early. That is a strong argument for beginning scope analysis during design development on negotiated work. Types 1, 2, and 5 require a reasonably complete set, since they depend on where obligations actually landed.
Is Type 7 worth spending time on?
It has the lowest priority of the ten, and it should still be checked, because the check is nearly free once you are already reconciling specifications against drawings in the other direction. The cost of Type 7 is rarely large. The irritation is real, and it lands on your project engineer during submittals.
How should we record gap types in our log?
A single type code field, one of ten values. It costs nothing per entry and it makes the closeout retrospective analytically useful, because you can aggregate by type across projects and see which detection method to strengthen. Without a type code, closeout produces anecdotes.
How can we find out what type of changes hurt us the most?
Pull the last three completed projects. Create a list of each scope change order, and associate a type for each order from Table 1. You will find that two or three types account for most of the dollars, and that is where to concentrate. Almost every team that runs this exercise is surprised by the result, usually by how much Type 5 dominates.
Does BIM coordination address any of these types?
Partially, and only Type 4 and Type 5, where geometric interfaces are involved. A federated model shows physical conflict. It does not represent division coverage, specification text, delegated design language, contractual boundaries, or subcontractor exclusions, which means seven of the ten types are outside what coordination can see. Coordination and scope analysis answer different questions and neither substitutes for the other.
How often should the taxonomy itself be revisited?
Rarely. The ten types are structural features of how construction documents and subcontract structures interact, and they have been stable for a long time. What should be revisited annually is the weighting: which types are generating your findings and your field surprises, and therefore where the effort should sit next year.
Expert Recommendations
Classify your last twenty scope change orders this month. It is the single most informative afternoon available to a preconstruction team, and the result will not be what you expect.
Build Types 3, 5, and 8 as standing queries rather than as review tasks. All three are searches, all three cover high-frequency categories, and searches do not suffer the compliance decay that review tasks do.
Make the Type 4 interface grid a hard gate on award. It is a policy decision costing nothing, and it prevents the category of dispute that most damages field relationships.
Get the boundary documents at intake, every project, without exception. Work letter, OFCI schedule, utility agreements, phasing plan. Type 9 is the type that ends up in litigation.
Add a type code field to your gap log. Ten values, one keystroke, and it converts your closeout retrospective from storytelling into analysis.
Stop reading document sets by discipline. This is the hardest recommendation here because it runs against how everyone was trained, and it is the one that addresses the highest-frequency gap type in commercial construction.
Tell your subcontractors which type of gap you are asking about. “This is an interface question, not a scope addition” changes the conversation completely, and it builds the kind of credibility that gets you honest numbers on the next bid.
Where the Taxonomy Leaves You
Ten types, and only one of them is the thing most people mean when they say scope gap. The other nine are the ones that cost money, and they cost money largely because they go unnamed. An unnamed failure mode cannot be searched for, cannot be assigned to a role, and cannot be counted at closeout.
Naming them is the whole trick. Once your team can say “that is a Type 5, run the cross-discipline query” or “that is a boundary gap, get the work letter,” scope review stops being a matter of experience and vigilance and becomes a matter of running ten checks. Vigilance fails at sheet 180. Checks do not.
Start by classifying what has already hurt you. The pattern will be clear, it will be specific to your market and your project types, and it will tell you exactly which two checks to build first.
