Most contractors we meet are profitable on paper and have never once asked whether their problem-solving counts as research. It usually does not occur to them. A framer spends three weeks working out why a wall assembly fails a moisture test in a 105-degree Washington County summer, tries four fixes, and moves on to the next job. The solution is built into the house, and the time is buried in labor cost. For some of those projects, that work is exactly what the federal research credit was written to reward.

The R&D tax credit for construction companies is real, it is underused, and it fails or succeeds on paperwork. Here is how the test works, what you need to keep, and how the rules look after the 2025 tax law changes.

Why contractors assume the research credit is not for them

The credit lives in Section 41 of the tax code, and the name sounds like a lab coat and a patent filing. In practice, the statute does not care what industry you are in. It cares about what your team did on a specific business component, such as a product, process, technique, or design.

For a builder, that business component might be a panelized wall system, a foundation design for a hillside lot, a new method for installing radiant barriers in an attic, or a prefabrication workflow that changes how you frame. Routine work built to a known plan does not qualify. Developing a new or improved approach, where the answer was not known when you started, can.

The credit is also not a deduction. A deduction lowers taxable income. A credit lowers your tax bill dollar for dollar, which is why even a modest claim gets a CPA’s attention.

Where the four-part test meets a St George job site

Every qualified activity has to pass all four parts of the test. The IRS describes the credit mechanics on its About Form 6765 page, and the four parts are worth walking through with a construction example.

Say a custom home builder in St George wants to cut cooling loads in a desert climate. Their idea is a site-assembled wall panel with a ventilated cavity and a reflective layer, finished in standard stucco.

| Test | What it requires | How the example meets it | |—|—|—| | Permitted purpose | The work aims at a new or improved function, performance, reliability, or quality | The panel is meant to improve thermal performance and moisture resistance | | Technological in nature | The work relies on engineering or physical science principles | Heat transfer, vapor diffusion, and structural loading drive the design | | Elimination of uncertainty | You did not know at the outset the capability, method, or appropriate design | The team could not tell whether the cavity depth would dry out or trap moisture under stucco | | Process of experimentation | You evaluated alternatives through modeling, simulation, or systematic trial and error | Three mock-up assemblies were built, instrumented, and tested before one went on a live job |

Notice what is absent: the project did not have to succeed. A failed mock-up that taught you something still counts as experimentation. What does not count is work after the design is final, cosmetic changes, or solving a problem by looking up the answer in a manufacturer’s spec sheet.

The statute also has a threshold: substantially all of the activities on a business component, generally read as 80 percent or more, must be experimentation for a qualified purpose. That is why we scope claims by component rather than by whole project.

A worked example with real numbers

This is a hypothetical, built to show the arithmetic. Our builder’s wall panel effort runs over about eight months.

  • Wages: a project manager, a lead carpenter, and two crew members spend a documented portion of their time on design, mock-ups, and testing. Qualified wages total $180,000.
  • Supplies: lumber, membranes, sensors, and stucco for the mock-ups total $40,000. These are consumed in the research, not equipment you depreciate.
  • Contract research: an outside structural engineer is paid $30,000 to model the assembly. Generally, 65 percent of that, or $19,500, counts.

Qualified research expenses come to $239,500. Under the alternative simplified method, the credit is 14 percent of the amount above half of your average prior three years of qualified spending. If that three-year average was $100,000, the base is $50,000, the excess is $189,500, and the credit is about $26,530. That is before any adjustment under Section 280C, where you either reduce your deduction by the credit or elect a reduced credit. We model both and pick the better one.

Utah also has its own research activities credit that builds on the federal definitions, so we check state eligibility in the same pass.

What counts as a qualified expense

There are three buckets, and construction businesses tend to trip on the second and third.

  • Wages. Employees doing the research, directly supervising it, or directly supporting it. The owner’s time counts if the owner is doing the work and is on payroll. Time spent bidding, estimating, or managing a crew on routine work does not.
  • Supplies. Tangible property used up in the research. Land and depreciable equipment are out. Test materials, mock-up lumber, and sensors are in.
  • Contract research. Generally 65 percent of what you pay an outside party, such as an engineer or testing lab, to perform qualified research on your behalf. You must bear the economic risk, and you generally need to retain rights to the results.

That last point matters for contractors. If a customer funds the work and you get paid regardless of whether it succeeds, the research may be treated as funded and fall out. Fixed-price work where you absorb the risk of failure looks very different from cost-plus work, so contract terms are part of the analysis. If you build for a developer or a government agency, read the contract language before assuming anything.

Documentation that holds up in an audit

This is where claims succeed or fail. The IRS does not need perfection, but it does expect records created while the work was happening, not a narrative assembled two years later.

What we ask construction clients to keep:

1. Project files. Design sketches, engineering calculations, RFIs, and change-order notes that show the uncertainty you faced. 2. Test records. Photos of mock-ups, moisture readings, blower-door results, and notes on what failed and why. 3. Time tracking. Hours by employee, by project, and by activity. A weekly code in your timekeeping system is far stronger than a year-end estimate. 4. Payments. Invoices from engineers and testing labs, with scope descriptions that match the research. 5. A short project summary. One page per business component stating the goal, the uncertainty, the alternatives tried, and the outcome.

This is easier when your books are already organized by job. If your job costing is solid, adding a research flag to a cost code costs almost nothing. If it is not, construction accounting is the place to start before any credit study.

A well-documented claim is also easier to defend. If the IRS sends questions, audit protection means someone on our team handles the response rather than you.

How Section 174 changes the math

Section 174 is the companion rule that determines how research costs are deducted. For tax years beginning in 2022 through 2024, businesses had to capitalize domestic research and experimental costs and amortize them over five years. That hurt cash flow even for companies that never claimed a credit, because wages for engineers and developers could no longer be deducted in full in the year paid.

The 2025 tax law (OBBBA) added Section 174A, which restores immediate expensing of domestic research costs for tax years beginning after December 31, 2024. Smaller businesses, generally those with average gross receipts of $31 million or less, may be able to apply the change retroactively to 2022 through 2024, subject to election deadlines. Foreign research is still amortized over 15 years.

Why this matters to a contractor: the credit and the deduction interact. Claiming the credit means adjusting your deduction or taking the reduced credit. Depending on your history, going back to amend earlier years may produce a refund. Whether that is worth the effort depends on your numbers, and we run it both ways before recommending anything. Our business tax planning work covers that modeling year-round, not just in March.

The payroll tax offset for newer companies

A smaller group of readers can use the credit before they owe any income tax. A qualified small business, generally one with under $5 million in gross receipts and no receipts more than five years back, can elect to apply up to $500,000 of the credit against payroll taxes. Up to $250,000 offsets the employer share of Social Security tax, and up to $250,000 offsets the employer Medicare share.

The election is made on Form 6765 with a timely filed return, and the offset is taken on Form 8974 against quarterly payroll filings. For a young St George contractor or design-build startup with a heavy payroll and little taxable income, that turns a credit that would otherwise carry forward into cash. Most established contractors will not qualify on the five-year test, but a company that spun out a new entity to develop a product line might.

Common Questions

Does a general contractor qualify for the R&D tax credit?

Yes, if the work meets the four-part test. A general contractor developing a new building method, assembly, or process can qualify, while routine construction to a standard plan does not. The claim is built around specific business components, such as a prefab system, and the hours spent experimenting on them.

How far back can we claim the credit?

You can generally amend returns for the prior three years, and unused credits carry forward 20 years. Because documentation drives the claim, older years are only worth pursuing if records from that time survive. We review what you have before estimating any refund.

Is the credit worth it for a small contractor?

It can be. In the example above, a $239,500 effort produced roughly $26,500 in credit. The cost-benefit depends on documentation quality and study fees. A smaller claim with clean records is often a better bet than a large one assembled from memory.

Find out whether your work qualifies

If you have developed a new building method, built test mock-ups, or paid an engineer to solve a problem the plans did not answer, bring us the project files and we will tell you honestly whether a credit study makes sense. Start a conversation on our contact page and we will review your last two years of projects with you.