The Real Cost of Uncontrolled RFIs in Design-Build Firms
Most coordination stops at clash detection.
If two elements don't touch, the model passes.
A model can be clash-free and still not constructible.
Over almost thirty years in AEC, running my own practice and working inside other firms' projects, I've learned to read the RFI log as a map of where the process let a decision slip.
The RFIs that come back most often aren't clashes. A shaft holds the ductwork but leaves no room to install or service it. A door fits the wall but loses the clearance it needs. Equipment fits the room but not its access path. Casework sits right where an MEP component has to go. Each discipline's drawings can be consistent on their own while the combined project isn't.
A second group has nothing to do with coordination. The drawings show what a condition should look like, but not how it gets built: a material transition without a detail, a penetration, an envelope termination, a wall-to-ceiling intersection nobody resolved. The contractor sees more than one way to read it and has to ask.
In design-build, both kinds of questions come back to the firm that drew the set, and the firm pays for them in schedule and senior hours first, and eventually in the owner's confidence. Each one is a control gap with a timestamp: the moment something that was detectable upstream reached the field.
Not every RFI can be prevented. Sites surprise people and owners change their minds. The ones worth going after are the preventable RFIs, the ones whose cause was already sitting in the documents or the model before construction started.
What it costs the firm
Two numbers help put this in front of a firm's leadership.
The largest study on RFIs, published by the Navigant Construction Forum in 2013, analyzed more than a million of them. Each one cost about $1,080 to review and answer. Projects between $5M and $50M, the range most small and mid-sized firms work in, generated 17.2 RFIs per million dollars, almost twice the overall average. On a $20M project, that's around 340 questions and roughly $370,000 in processing alone, before any rework. The study is more than ten years old and labor costs have risen since, so these figures are best read as a floor.
The second number comes from the people who pay when errors escalate. In Ames & Gough's 2026 survey of 15 leading A/E professional liability insurers, 60% reported higher claim severity in 2025, up from 41% in 2023, and 82% paid multimillion-dollar claims. The survey doesn't break claims down by cause. For a firm leader, the direction is still clear: every document conflict that reaches the field is a potential claim, and claims are getting more expensive.
Seven control points that close the gap
Most firms already do some of these informally. The difference is doing them on every project, with someone accountable. Together they check two things before a set goes out: that the information is coordinated, and that it's complete enough to build from.
Seven control points across the project
-
1Decision logSD · DD
-
2Interface ownersSD · DD · CD · CA
-
3Phase gatesEnd of SD · DD · CD
-
4Critical dimensions reviewDD · CD
-
5Detail completeness reviewDD · CD
-
6Validate before sharingEach phase boundary
-
7RFI → rule loopCA ↺ SD · next project
Source: FlowBIM.ai
1. Close decisions before they reach the model
Many RFIs start as a decision that was never formally made. Someone models a placeholder, and the placeholder quietly becomes the answer.
A decision log fixes this with very little overhead. Each open decision gets an owner, a due date, and the phase by which it must be closed. Schematic design closes massing, program, and major systems. Design development closes assemblies, door and hardware types, and ceiling strategy. Anything still open when construction documents start gets flagged instead of buried.
The log also protects the firm later. When an owner-driven change shows up as an RFI, there is a record of when the decision was made and by whom.
2. An owner for every interface
Clearances are a good example of a problem that belongs to no one. The architect lays out the room, the engineer sizes the equipment, and the space between them is nobody's job until the contractor asks about it.
A responsibility matrix for the few interfaces that generate most RFIs changes that: shafts and risers, doors and their clearances, ceiling zones against ductwork and structure, casework against MEP. The matrix doesn't need to be long. What it needs is a person's name on every line.
3. Gates at the end of each phase
A coordination gate is a short list of conditions a set must meet before it moves forward. At the end of design development, for example: no open clashes in corridors and shafts, no "TBD" in door, window, or finish schedules, and the critical dimensions review (point 4) closed.
If the conditions aren't met, the set doesn't advance, or it advances with the open items listed and assigned. Moving forward with known gaps is sometimes the right call. It should be a decision someone makes on purpose. Too often it's something the framer finds out first.
4. A critical dimensions and clearances review
Clash detection answers one question: do two objects occupy the same space? It doesn't check minimum clearances, maintenance zones, installation tolerances, code-required dimensions, or equipment access.
Before each major CD or IFC issue, critical conditions get reviewed across disciplines instead of sheet by sheet: shafts and risers, doors and swings, accessibility clearances, corridors and egress paths, equipment rooms and access, ceiling zones, casework against MEP. Each item gets one line:
Requirement → Designed dimension → Source → Discipline → Status
That line moves QA/QC from "does it look right" to "does it meet the requirement." Clearance zones built into the families and rule-based model checks can take over much of the repetitive part.
Coordinated?
- Dimensions
- Clearances and access
- Across disciplines
Critical dimensions review
Complete?
- Drawings
- Details
- Specifications
Detail completeness review
Constructible
Source: FlowBIM.ai
5. A constructability and detail completeness review
A drawing set can look complete sheet by sheet and still leave conditions unresolved for the person who has to build them. This review asks one question of every critical condition:
Could someone build this from the information provided without having to ask us another question?
It focuses where the answer is most often no: material transitions, envelope interfaces, floor, wall and ceiling intersections, openings and penetrations, waterproofing and flashing, terminations and attachments. For each one, the check runs in order: condition identified, detail exists, detail referenced correctly, dimensions complete, materials defined, adjacent systems coordinated.
Much of this cross-checking between plans, sections, details, and specifications can be automated. What AI can and can't check is the subject of How AI Is Changing Pre-Construction Validation in AEC.
6. Validate before information is shared
ISO 19650 formalizes a simple habit: information is checked before it moves from work in progress to shared, where it becomes someone else's input. In a small firm, that can be a checklist and a sign-off before a model or sheet set goes to consultants or to the field. The habit matters most in deadline weeks, which is exactly when it tends to get skipped.
7. Turn every recurring RFI into a rule
Every RFI that still gets through says something about the process. The useful sequence is short: RFI, root cause, control, rule, prevention. Coding each RFI by origin, discipline, and the phase where it should have been caught shows which ones keep coming back.
Those recurring causes become checks in the next project's gates and reviews, so the next project starts with the lesson already built in.
One line per critical condition
Requirement → Designed dimension → Source → Discipline → Status
| Condition | Requirement | Designed dimension | Source | Discipline | Status |
|---|---|---|---|---|---|
| Mechanical shaft, level 3 | Service clearance for duct access | [value] | Equipment data / MEP | MEP ↔ Architecture | Open |
| Corridor door on accessible route | Required maneuvering clearance | [value] | Accessibility code | Architecture | Resolved |
| Rooftop unit access path | Clear path for replacement | [value] | Manufacturer data | MEP ↔ Structure | Open |
Source: FlowBIM.ai
Where to start
A firm of 10 or 30 people doesn't need all of this at once, and the first month doesn't require buying anything or hiring anyone.
What it requires is a leader who asks for it:
1. Code the RFIs from one recently finished project: coordination, completeness, or other.
2. Pick the three causes that repeat most.
3. Add one checkpoint for each, at the phase where it should have been caught.
4. Track two ratios on the next project:
Pre-issue detection rate: coordination issues found before issue ÷ all coordination issues.
Document completeness RFI rate: RFIs caused by missing or ambiguous information ÷ all RFIs.
If the first ratio goes up and the second goes down, detection is moving upstream.
What changes
Clash detection keeps its place. It just isn't the finish line. In design-build, a good share of RFIs is the firm's own process arriving late, and the checks that catch them look at requirements and completeness. Processing RFIs faster helps the schedule a little. Building those checks into the process is what eventually shrinks the log.
I've dealt with those RFIs from both sides of the table, as the one who drew the set and as the one helping another firm untangle it.
The ones that hurt most were usually easy questions that nobody had checked before the set went out.
If your RFI log looks familiar, the first step is finding where control breaks down. Start with a Production Control Diagnostic™.
Frequently asked questions
Why wasn't this caught before the drawings were issued? Because most model coordination looks for geometric clashes, while many construction RFIs come from dimensions, clearances, and relationships that need explicit verification. A model can pass clash detection and still not be buildable as drawn.
How can a complete CD set still generate RFIs for missing information? Every expected sheet can be in the set while individual conditions stay unresolved or open to interpretation. The contractor finds those gaps when planning installation and sequencing.
A completeness review is meant to find them first.
Can RFIs be eliminated entirely? No. Differing site conditions and owner changes will always produce legitimate questions. The target is preventable RFIs: the ones whose cause was already detectable in the documents or the model before construction.
Why do RFIs cost a design-build firm more? The design-build entity holds both design and construction responsibility. When a document gap causes an RFI, there's no separate designer to allocate the cost to through a change order, so the firm absorbs the cost and the hit to its reputation.
Does reducing RFIs require new software? Not at first. A decision log, a responsibility matrix, and phase gates cost nothing but discipline. Model checking and AI-assisted validation pay off once those control points exist. Without them, software just speeds up the same gaps.
References
• Navigant Construction Forum. Impact & Control of RFIs on Construction Projects. April 2013.
• Ames & Gough. 2026 A/E Professional Liability Survey Results. 2026.
• ISO 19650 series: information management using BIM, including the work-in-progress to shared transition.
Stop RFIs before they reach the field.
The control points in this article aren't theory. We install them in your firm's workflow.