The $15,000 Change Order That Should Have Been an RFI

The electrical subcontractor installs 400 linear feet of conduit in the ceiling plenum. The mechanical sub shows up to install ductwork and the conduit is in the way. The electrical says the routing was per the drawings. The mechanical says the duct routing was per the drawings too. They're both right. The drawings conflict.

Nobody caught it. Nobody submitted an RFI asking which routing takes priority. Now you have a change order: $15,000 to reroute the electrical conduit, plus two days of schedule impact. The architect argues it was the GC's responsibility to coordinate. The GC argues the drawings should have been coordinated before they were issued. The owner pays the bill and loses trust in both of you.

This happens on every project. Not this exact scenario, but some version of it. A conflict or ambiguity in the contract documents that was identifiable during preconstruction or early construction, that nobody questioned, that turned into field rework and a change order. The FMI/CURT study puts the industry cost of rework from miscommunication at $31 billion per year. A meaningful percentage of that is change orders that should have been RFIs.

The three types of avoidable change orders

Drawing conflicts

The structural drawings show a beam at one elevation. The mechanical drawings show a duct at the same elevation. The architectural reflected ceiling plan shows a soffit that doesn't have room for either. Three disciplines, three different assumptions about the same space. The 3D coordination model might catch it, if the model is up to date and if someone is actually reviewing the clash reports. On projects without full BIM coordination, these conflicts hide until installation.

The RFI that prevents this: “At grid C-4, second floor, the structural beam (S-201) and the HVAC supply duct (M-301) are shown at the same elevation. Which routing takes priority? Mechanical reroute requires 4 additional hangers and affects the return air path.” That RFI costs zero dollars and takes 10 minutes with the right tool. The change order costs $15,000 and takes two weeks.

Specification ambiguity

The spec says “provide moisture protection at all below-grade concrete.” Does that mean a sheet membrane? A spray-applied coating? A crystalline admixture in the concrete mix? The spec references three sections, each describing a different method. The estimator priced the cheapest option. The architect meant the most expensive one. The delta is a change order.

This is the most common source of avoidable change orders in my experience. Specifications are written by multiple authors. Section 03 30 00 (concrete) was written by the structural engineer. Section 07 10 00 (waterproofing) was written by the building envelope consultant. Section 07 11 00 (dampproofing) was written by the architect. They don't always agree. The GC who reads all three and submits an RFI asking “which method is intended?” avoids the change order. The GC who prices the cheapest interpretation gets the job and then gets the change order.

Scope gaps between trades

Who installs the blocking for the handrail? The framing sub says it's not in their scope. The railing sub says they install the rail, not the backing. The spec says “provide backing as required for attachment.” Required by whom? The framing sub read “provide” as “someone else provides.” The railing sub read it the same way.

Scope gaps live in the white space between trade scopes. They're invisible during buyout because each sub scopes to their own spec section. The gap only appears when someone needs the blocking and it's not there. An RFI during preconstruction asking “Section 06 10 00 does not include blocking for Section 05 52 00 handrail attachment. Please confirm which trade is responsible” closes the gap before it becomes a problem.

Why the RFIs don't get written

The answer is always the same: time. The project engineer is managing 200 submittals, 50 open RFIs, daily reports, and coordination meetings. Reading the full specification during preconstruction, cross-referencing it against the drawings, and identifying conflicts and ambiguities is work that takes hours. It's high-value work, but it competes with urgent work. The urgent work wins every time.

By the time the conflict is discovered in the field, the RFI is reactive instead of proactive. The rework is already underway or unavoidable. The change order is the only path forward.

Building a proactive RFI culture

The GCs with the lowest change order rates on their projects have one thing in common: they write more RFIs, earlier. Not nuisance RFIs. Targeted questions about real conflicts and ambiguities identified during preconstruction review.

• Assign a preconstruction RFI review to the PE before the first subcontract is awarded. Read the spec cover to cover. Flag every ambiguity, conflict, or scope gap. Submit RFIs on the critical ones.
• During buyout, require each sub to identify conflicts between their scope and adjacent trades. Put it in the subcontract: “Subcontractor shall review contract documents and submit RFIs for any conflicts or ambiguities prior to commencement of work.”
• Track the ratio of proactive RFIs (submitted before work starts) to reactive RFIs (submitted after a conflict is discovered). A healthy project should be 60/40 or better.

Where AI changes the equation

The preconstruction spec review that nobody has time for? An AI tool can do a first pass in minutes. Upload the spec. Ask it to identify conflicts between sections, ambiguous scope language, and gaps between trade scopes. It won't catch everything, but it will catch the obvious ones that a busy PE would miss under time pressure.

The math is straightforward. One $15,000 change order avoided pays for a year of AI tooling. Most projects have a dozen of these hiding in the documents.

Tim Lewis spent 25 years in commercial construction, including a decade as Regional Director at Harper General Contractors, a $500M ENR Top 400 firm. He founded Contractor-AI to help construction companies implement AI where it delivers real ROI.