RFI in Construction: The Question That Costs a Thousand Dollars to Ask

Ask what an RFI is and every guide gives you the same line: a Request for Information, the form a contractor files when the construction documents don't say enough. True, and dangerously incomplete. An RFI is the moment a contradiction that has been sitting in the documents since issue day finally surfaces in the field, with a crew standing next to it. What the form says matters less than what the answer becomes - and that is the question almost none of the guides answer.

What an RFI actually is

A Request for Information is a formal, logged question from the party building the work to the party who designed or commissioned it. The contractor asks; the architect, engineer, or owner's representative answers; the exchange goes into the RFI log with dates attached. That log is evidence in the making: on most projects it will be read again, months later, by someone deciding who owes whom money.

The trigger is almost always the same. The issued for construction set froze the design at a moment in time, contradictions included. A dimension on the structural sheets disagrees with the architectural ones. A spec section calls for a product the drawings never located. A detail references a sheet that doesn't exist. The field cannot pour, hang, or route around an ambiguity, so the work stops being built and starts being asked about.

An RFI is not a submittal, which flows the other way - the contractor proposing what they intend to install. And it is not a change order, although the distance between the two is shorter than most people managing the log believe.

What one question costs

The numbers on this are old but nobody has published better ones, which itself says something. A widely cited 2013 Navigant Construction Forum analysis put the average cost of processing a single RFI at roughly $1,080, with around eight hours of design-side time to receive, review, and answer each one. Industry rules of thumb run 10 to 15 RFIs per million dollars of project value, and large commercial jobs commonly log several hundred. The same analysis found that more than one RFI in five never receives an answer at all - in the Americas, roughly one in four.

Response times make it worse. Contracts typically allow 7 to 14 days for a response; the same 2013 data put the median at just under 10. While the question routes, the crew that hit the ambiguity is resequenced or standing still, and standing still on a commercial site bills by the day. That is the arithmetic under the paperwork: the RFI is one of the most expensive ways to discover a document contradiction, because it is discovery at field prices, one contradiction at a time.

The question nobody answers: what is the answer, contractually?

Every guide explains how to write a clear RFI. Almost none of them tells you what the response actually is. The honest answer: your contract decides, and many contracts decide differently.

Broadly, an RFI answer lands in one of two places:

A clarification. The answer interprets what the documents already required. No new scope, no new money, no new time. Most answers claim to be this.

A directed change. The answer tells the contractor to do something the issued documents did not require - a different product, a heavier member, a relocated run. That is new scope wearing a clarification's clothes - what claims people call a constructive change when it never gets formalized - and it is supposed to become a change order with a price and a schedule adjustment attached.

The fights start in the gap between those two, because the answer rarely announces which one it is. A contractor who builds a directed change without flagging it has, under most contracts, performed extra work for free. An owner who lets "clarifications" quietly accumulate scope has an unpriced project. This is why experienced teams read RFI answers the way lawyers read letters: not "what does this tell me to do" but "what does this make me owe."

Two disciplines follow from that, one per side of the table:

If you are asking: put your proposed interpretation in the RFI itself, with a cost and schedule note if the answer could move either. An RFI that says "we assume detail 5/S301 governs; please confirm" converts a fishing expedition into a record. If the answer directs new work, say so in writing before you build it.

If you are answering: answer the question that was asked, in the documents' own language, and state plainly when the response changes scope. An ambiguous answer does not protect the design team; it moves the dispute to closeout, where it is supposed to survive onto the as-built record and rarely fully does.

The log is a confession

Step back from any single RFI and read the log as one document - five hundred questions on a large job. Most of them mark a place the issued set did not agree with itself or with the site. The log is the project's own record of what was, in large part, discoverable on issue day - found instead at field prices over eighteen months.

That reframing matters because most of the industry treats RFI volume as weather - something that happens to projects. The management guides optimize the routing: better forms, faster workflows, cleaner logs. All real improvements, and all of them accept the premise that the contradiction gets found by a foreman with the work open in front of him.

Some RFIs genuinely cannot be prevented - field conditions, owner changes, things no document could have known. But the subset born from the documents disagreeing with each other existed, findable, before the first one was filed. Catching drawing-spec contradictions before the set is issued is the only version of this work that happens at desk prices.

Where Lintel fits

Lintel reads the full document set - drawings, specs, schedules - and surfaces where the documents disagree with each other. The contradictions that become RFIs are exactly that shape: the dimension that differs between disciplines, the spec section the drawings contradict, the detail that references a sheet that isn't there. AI plan review runs that reading before issue, so the question can be asked and answered while it is still a markup between professionals, not a logged, dated, thousand-dollar entry with a crew attached.

The difference is when the contradiction is found: by your team the week the set is assembled, or by the field in month four, one RFI at a time.

Common questions

What does RFI stand for in construction? Request for Information - a logged, formal question from the builder to the design or ownership side about what the contract documents require. It is distinct from RFP, RFQ, and RFT, which are procurement documents used before a contract exists.

Is an RFI a change order? No, but it is often the first page of one. If the answer only interprets the existing documents, nothing changes. If it directs work the issued documents did not require, it should be converted into a change order with cost and time attached - and the contractor should flag that before building it, not after.

How long does an RFI response take? Contracts commonly allow 7 to 14 days; the widely cited 2013 analysis put the median response at just under 10 days. The cost is rarely the response itself - it is what the crew that raised the question does while the answer routes.

See what Lintel finds in your document set before it becomes an RFI log - Book a demo