The RFI Your Team Answered Three Times This Year

What Does It Actually Cost to Answer an RFI in Construction?

Answering a single RFI in construction costs an average of $1,080 in administrative and technical review time¹. On a typical project, that adds up to nearly $860,000— before a single nail is driven or beam is set.

Per a widely-cited 2013 study from the Navigant Construction Forum, conducted in partnership with CMAA (Construction Management Association of America) across more than 1,300 construction projects:

These aren't abstract figures. On a $50M project, you're looking at 500–750 RFIs⁸ and a process overhead that could rival a small subcontract. And that 22% no-response figure? The Navigant study puts it plainly: "The no response rate is one of the major indicators that a dispute will arise."¹

The $31.3 billion industry-wide figure from the PlanGrid/FMI Construction Disconnected report² puts the RFI problem in broader context. Poor communication accounts for 26% of US construction rework; incomplete or inaccurate project information accounts for another 22%². RFIs are one visible symptom of a much larger information problem.

But volume is only part of the problem. The deeper issue is which RFIs are being submitted at all.

Why Do Construction Teams Keep Answering the Same RFIs?

About 13% of construction RFIs ask questions that were already answered somewhere in the project documents¹. The answer existed— it just wasn't findable. That's one problem. The other is that even when an RFI does get answered well, that answer rarely survives past project closeout.

There are two distinct failure modes feeding this:

• Retrieval failure (within-project): The answer is in the documents, but no one can find it in time. Documents on a complex project can run into the thousands of pages. This isn't about people not trying— it's that the search infrastructure doesn't exist.
• Transfer failure (cross-project): The answer was found once— on a different project, two years ago— but it never made it into any system the next team could access.

The within-project problem has a price tag. That 13.2% of unjustifiable RFIs translates to an average of $113,400 per project in preventable administrative overhead¹. That's a check your firm is writing every time someone submits a request asking a question that's already in the contract documents.

The cross-project problem is harder to quantify but arguably worse. Peer-reviewed research in Expert Systems with Applications³ describes it as "organizational amnesia"— construction firms "cannot systematically transfer knowledge" because it ends up in "informational limbo" when project teams disband. Knowledge lives in individual email threads, project folders, and people's heads. When the project closes, it scatters.

If you've worked in construction for more than a couple of years, you've felt this. No study needed. The Navigant data just puts a number on something you already know is true.

The Harder Truth: Are Some RFIs Just Waste?

From a lean construction perspective, an RFI isn't just process overhead— it's a signal of design failure. The lean framing is blunt:

"An RFI is not a request for information. It is rework for information."⁴ — Elevate Constructionist

They're not wrong. RFIs that could have been prevented through better design coordination, earlier specialty contractor involvement, or direct communication channels represent upstream failures. High RFI volume is a design quality signal— not a workflow problem to optimize around.

But the lean critique isn't complete. Complex projects will always generate some legitimate RFIs— ambiguity in drawings, field conditions that don't match the spec, decisions that weren't made at design and got punted to construction. The goal isn't zero RFIs. It's zero avoidable ones.

The practical takeaway: high RFI volume is a design quality signal— not a workflow problem to optimize around. It tells you where design coordination broke down before field work started. Most teams aren't in a position to redesign their entire design coordination process before the next project starts. They need a way to handle the RFIs they have— and prevent the avoidable ones. That's exactly where AI has started to change the math.

AI and the Construction RFI: Prevention Matters More Than Speed

AI tools are now embedded in the major construction management platforms— and the most important thing they can do isn't answer an RFI faster. The real job is checking whether the question has already been answered before the RFI gets submitted at all.

AI as faster response. Procore's RFI Creation Agent⁵ automatically scans project documents and surfaces answers, reducing turnaround from days to seconds for document-answerable questions. Procore Assist provides conversational AI for contextual answers across project documents, specs, and building codes. Procore AI also includes schedule impact analysis features that assess the likelihood of delay when a new RFI comes in. That's meaningful acceleration— but you're still paying $1,080 each time a new RFI gets submitted.

AI as prevention— this is where the real leverage is. Platforms like Datagrid⁶ check whether similar questions already exist before a new RFI is formally submitted. Pattern recognition groups recurring questions across project history, both within the current job and across the firm's closed projects. This directly addresses both failure modes: retrieval failure within a project and transfer failure across projects.

The distinction that matters: AI that speeds up response still costs you $1,080. AI that prevents the question from being asked saves you $1,080.

Platform selection depends on your current stack. Here's the breakdown:

• Procore: RFI Creation Agent (document scanning and answer surfacing), Assist (conversational AI across project docs), and schedule impact analysis features
• Autodesk Construction Cloud (ACC): Document management with AI-powered search (the platform formerly known as BIM 360)
• Datagrid: Duplicate detection before submission, pattern recognition across project history (cross-project historical analysis requires data integration setup)
• Civils.ai: AI-assisted RFI workflow for AEC project management

Which tool makes sense depends on where you are. If you're already in Procore, start with what you have before adding a new vendor. If you're evaluating fresh, ACC and Datagrid serve different scale and integration needs— worth a product conversation with each before committing.

Industry estimates suggest AI-driven RFI workflows can reduce preparation and response time by 50–70%⁷— though that figure comes from Varseno's industry reporting, not primary research, so treat it as directional rather than a target. Understanding what an AI agent actually does helps explain why the prevention use case is now technically feasible where it wasn't before.

One honest caveat: AI tools are only as good as the documentation they're trained on. If project documents are inconsistently organized or tagged across five different folder structures, AI can't help with retrieval. Good documentation hygiene is a prerequisite— not an afterthought. Just because it's easy to deploy an AI tool doesn't mean it'll work. The foundation matters.

A Practical Starting Point for Mid-Market AEC Firms

Getting RFI volume under control doesn't require replacing your project management platform or implementing a firm-wide AI strategy in month one. It starts with two decisions: where to capture what you know, and when to check what you already have.

Here's what firms getting ahead of this tend to do:

1. Audit what you have. Most AEC firms have years of RFI archives in Procore or ACC. That's a knowledge base— but only if it's accessible. Before adding any AI tools, confirm your historical RFIs are organized, tagged, and searchable. If they're not, start there.

2. Change the submission habit. Before submitting an RFI, require a search step. Does this question already exist on this project? Has a similar spec come up on past jobs? This low-tech process change can catch a meaningful share of that 13% before they become formal requests. Automating the search step is straightforward once the habit is in place.

3. Choose the right AI level. For smaller AEC firms, start with the AI features already in your existing platform (Procore AI⁵, ACC's AI search). For larger firms or those with custom workflows, tools like Datagrid⁶ offer deeper pattern recognition across your full project history. Use 10–15 RFIs per $1M of project value⁸ as your benchmark— if you're tracking significantly higher, the retrieval and transfer failures are compounding.

4. Treat closed-project archives as assets. Build a post-project knowledge capture step— a short document summarizing the most common RFIs and their resolutions. Low-tech. High-leverage. The change management required to make this habit stick is often harder than the technical setup, but it's the part that compounds over time.

5. Measure what you're managing. Track RFI volume per $1M of project value. Track your no-response rate. These metrics already exist in your platform— use them. If your no-response rate is climbing, that's an early dispute indicator.

The question isn't whether to use AI for RFIs— it's whether to deploy it as a speed tool or as a memory system. The firms who are actually getting ahead of this are choosing the second one.

Common Questions About RFIs in Construction

What is an RFI in construction?

An RFI (Request for Information) is a formal written request from a contractor or subcontractor to the project's design team— typically the architect or engineer— asking for clarification on drawings, specifications, or site conditions. It's a standard part of the construction process, used to resolve ambiguities before field work proceeds.

How long does an RFI take to answer?

The median RFI response time is 9.7 days, per Navigant Construction Forum research¹. Contract standards typically set 5–14 days depending on the complexity of the question. Delays beyond that timeline create schedule risk— and unanswered RFIs are a documented leading indicator of disputes.

How much do RFIs cost a construction project?

Answering a single RFI costs an average of $1,080 in administrative and technical review time¹. On a typical project generating roughly 800 RFIs, that adds up to nearly $860,000 in overhead— before any schedule impact is counted.

What percentage of RFIs are unnecessary?

About 13% of RFIs ask questions that were already answered somewhere in the project's contract documents¹. These preventable requests cost an average of $113,400 per project¹. That's not an incompetence problem— it's a search and retrieval problem.

Can AI reduce construction RFI volume?

Yes— AI tools embedded in platforms like Procore⁵ and Datagrid⁶ can now detect duplicate questions before an RFI is formally submitted, and surface document-based answers in seconds. Industry estimates suggest AI workflows can reduce RFI preparation and response time by 50–70%⁷, though independent benchmarks are limited.

What happens when an RFI goes unanswered?

According to the Navigant Construction Forum study¹, 22% of all construction RFIs receive no response. The research notes that the no-response rate "is one of the major indicators that a dispute will arise." Monitoring your no-response rate is early warning for legal exposure.

Where to Start

The RFI problem isn't going away— but it is solvable. The path forward is treating your RFI archive as institutional memory rather than administrative overhead.

Answering the same RFI twice (or three times) is a knowledge management failure. The AI tools available now address this at the infrastructure level— not just the speed level. Firms using them for prevention, not just response, are changing what information management actually costs— on every project they run.

RFI volume isn't just a process metric— it's a measure of how well your firm's knowledge survives from one project to the next.

If you're evaluating where AI fits in your construction workflows— not just for RFIs but across project operations— that's a strategic decision worth getting right. If that's the conversation you're trying to have— what AI is actually ready to do on your projects, versus what's being oversold— an experienced AI implementation partner is worth talking to. The goal isn't more RFIs handled faster. It's fewer RFIs because the knowledge lives somewhere accessible.

FAQ

If AI speeds up RFI responses, why doesn't that solve the cost problem?

Speed tools reduce turnaround time, but the $1,080 administrative and technical review cost per RFI is incurred the moment a new RFI is formally submitted—not when it gets answered. AI that prevents a question from being asked in the first place saves $1,080; AI that answers it faster still costs $1,080. That's why the article identifies prevention as the higher-leverage use case.

What's the difference between retrieval failure and transfer failure?

Retrieval failure happens within a single project—the answer exists somewhere in the contract documents, but no one can locate it fast enough, which accounts for roughly 13% of all RFIs and $113,400 in preventable overhead per project. Transfer failure happens across projects—an answer was found on a previous job but never made it into any system the next team could access, a pattern researchers describe as "organizational amnesia." Both failure modes drive repeated RFI submissions; they just operate at different scales.

What should a firm do before adding any AI tools to its RFI workflow?

The article is explicit that AI is only as good as the documentation it works from. Before deploying any tool, firms should confirm that historical RFIs in Procore or ACC are organized, tagged, and searchable—because inconsistently structured archives prevent AI from helping with retrieval at all. Good documentation hygiene is described as a prerequisite, not an afterthought.

How do I know if my project's RFI volume is a problem worth measuring?

The article cites 10–15 RFIs per $1M of project value as an industry benchmark—tracking significantly above that range suggests retrieval and transfer failures are compounding. Equally important is the no-response rate: the Navigant study found 22% of RFIs receive no response, and identifies that figure as a leading indicator of disputes. Both metrics are already available inside standard project management platforms.

References

1. Navigant Construction Forum / CMAA, "Impact & Control of RFIs on Construction Projects" (2013) — https://www.cmaanet.org/sites/default/files/resource/Impact%20&%20Control%20of%20RFIs%20on%20Construction%20Projects.pdf
2. FMI Corporation / PlanGrid, "Construction Disconnected" (2018) — https://www.autodesk.com/blogs/construction/survey-plangrid-fmi/
3. ScienceDirect / Expert Systems with Applications, "A novel framework of knowledge transfer system for construction projects based on knowledge graph and transfer learning" (2022) — https://www.sciencedirect.com/science/article/abs/pii/S0957417422003906
4. Elevate Constructionist, "The Truth About RFIs" — https://elevateconstructionist.com/the-truth-about-rfis/
5. Procore Technologies, "Procore Launches Procore AI with New Agents to Boost Construction Management Efficiency" (2024) — https://www.procore.com/press/procore-launches-procore-ai-with-new-agents-to-boost-construction-management-efficiency
6. Datagrid, "How AI Agents Automate Construction RFI Workflow" (2024) — https://datagrid.com/blog/ai-agents-improve-rfi-process
7. Varseno, "How AI Is Transforming Construction RFI & Submittals in 2026" (2025) — https://www.varseno.com/ai-transforming-construction-rfi-and-submittals/
8. Procore, "RFIs: A Contractor's Guide to Requests for Information" (2024) — https://www.procore.com/library/rfi-construction