Harper didn’t approve R&D credits — it made weak claims more dangerous. Here’s what it actually means, and why the firms that win audits build their defense before they file.
R&D Tax Intelligence
February 23, 2026
Many design-build and construction firms now cite Harper v. Commissioner as proof that R&D credits are available for design and construction work. That reading is wrong — and it is becoming dangerous.
The Harper case involved approximately $46,656 and $778,610 in claimed R&D credits across two tax years. What the Tax Court decided was narrow. It did not approve the credits. It did not validate the taxpayer’s methodology. It did not lower the evidentiary bar for qualification. Harper addressed a procedural failure by the IRS — not substantive eligibility.
Understanding that distinction is the difference between a defensible credit and a costly unwind.
In Harper, the IRS sought summary judgment by arguing that design-build activities could never satisfy the “business component” requirement under Section 41. The court rejected that categorical position. It held that the IRS could not disqualify the claim without examining the taxpayer’s actual activities and evidence.
That was the entirety of the ruling. The court did not decide whether the research qualified. It did not rule that the taxpayer met the four-part test. It did not resolve ownership, wage allocation, or substantiation issues. It allowed the case to continue. Harper forced IRS analysis before any decline — it did not deliver taxpayer approval.
This is where misinterpretation sets in.
The burden of proof remained with the taxpayer. The qualification standard did not soften. The taxpayer still must demonstrate technical uncertainty, experimentation, and qualified purpose through credible, project-level evidence. Harper did not excuse weak documentation or post hoc reconstruction.
Surviving summary judgment does not equate to surviving audit, trial, or appeal. Harper preserved the process — it did not predetermine the outcome. And the delay it creates can make things worse. When a claim is filed without a disciplined record, “support” often gets built later in the most dangerous way possible: rushed, reconstructed, and inconsistent. That is exactly the pattern IRS exam teams and courts punish most severely.
Here is the uncomfortable truth. Harper effectively guarantees that if the IRS cannot win on a blanket legal argument, they will win — or adjust — by testing your facts. That is not a hypothetical. That is the current posture of R&D exam teams nationwide.
In modern R&D exams, “facts” means two things: complete, traceable data — your full project universe tied directly to claimed QREs — and a methodology that holds up, especially when sampling is used.
For design-build and construction firms managing many projects, sampling is often necessary. But sampling without discipline is where claims die.
The IRS will review facts. You must be ready with facts — and those facts must have been built before the exam began, not during it.
Harper should be read as a correction, not a green light. It prevents the IRS from skipping factual analysis, but it leaves the evidentiary burden intact. Firms relying on Harper as a substitute for disciplined documentation are relying on procedure to overcome substance.
National Tax Group & TaxDrone.AI
This is the part no competitor wants to say out loud. Most R&D credit providers — consultants, CPA firms, credit mills — produce studies that are adequate at filing. The problem surfaces at exam, where adequacy at filing and defensibility under scrutiny are two very different standards.
The gap between a filed credit and a proven credit is where most adjustments happen. If your current or prior provider did any of the following, your claim carries real exposure:
This is what the difference looks like when an IRS exam actually arrives.
What the IRS Tests
NTG + TaxDrone.AI
Documentation timing
Captured in real time, project by project — never reconstructed
Business component mapping
Each claim mapped to an IRS-defined business component with evidence
Sampling methodology
Statistically defensible, Rev. Proc. 2011-42 compliant, fully documented population
Wage/QRE allocation
Costs aligned to specific, documented research activities — not titles or estimates
Non-qualifying activity
Flagged and excluded early — never enters the claim in the first place
Current case law alignment
Continuously updated against IRS guidance and court decisions — including Harper
Most firms think about audit defense when the IRS letter arrives. By then, the file is what it is. National Tax Group and TaxDrone.AI operate on a fundamentally different principle: we document the credit the way it must be defended — from the start — so if an IRS review comes, the file is already built for exactly that moment.
That is not incremental improvement over how other providers work. It is a different process entirely.
We document R&D credits the way they must be defended — from the start — so the IRS review, if it comes, finds a file that was built for exactly that moment.
Every engagement begins with a rigorous eligibility filter against current case law — including Harper. Activities that cannot withstand factual scrutiny are excluded before a credit is ever claimed. The most expensive R&D credit is the one you “win” today and unwind later with adjustments, penalties, interest, and professional fees. We eliminate that outcome at the source.
TaxDrone.AI captures evidence as work happens — not reconstructed after year-end, not assembled under the pressure of an IDR. The platform continuously checks against evolving IRS guidance so your substantiation standards keep pace with what courts are actually enforcing right now.
TaxDrone.AI is not a generic tax tool. It was built by National Tax Group specifically to solve the documentation and substantiation failures that cause R&D credits to fail under IRS review.
Scans your full project data and applies IRS-approved statistical sampling to identify more qualifying expenses than manual review — with fewer errors.
Every business component is evaluated automatically against Section 41's four-part test — including automated shrink-back analysis when components fail at the whole level.
Population defined, stratified, and randomly sampled using IRS-compliant randomizers. Projections are statistically valid — not estimated, not assumed.
Built by experienced tax experts. Every study produces airtight substantiation designed to withstand IRS scrutiny — no weak spots, no shortcuts.
Comprehensive documentation for both state and federal R&D tax credit returns — fully prepared and consistent across jurisdictions.
Clients consistently identify 2–3× more credits than with other providers — because the platform finds qualifying expenses that manual reviews miss.
National Tax Group & TaxDrone.AI — R&D Tax Intelligence
Harper did not make R&D credits easier. It removed a shortcut the IRS should never have used — and in doing so, it made the facts the only battleground. If your documentation is weak, Harper just guaranteed the IRS has to find that out the hard way, in an exam where the costs fall on you.
If you are relying on Harper to justify a claim, you are relying on procedure instead of proof. If you used another provider and are uncertain whether your file was built to the standard exams actually require, that question is worth answering before the IRS asks it for you.
Complete data. Statistically supportable sampling. Documentation standards designed to meet and exceed IRS and court expectations. See the process that makes “never lose an audit” a real outcome — not a marketing claim.
