Many of your clients qualify for the R&D tax credit and don't know it. Here's a practical guide for accountants on which industries, activities, and companies are eligible.
Disclaimer: This article is for educational purposes only and is not tax or legal advice. Talk to your tax advisor about your company’s specific situation.
The Tax Court Just Asked the Question Every Founder Should Ask About R&D Credits
There is a line in a recent U.S. Tax Court opinion that should make every startup founder pause before filing or amending an R&D tax credit claim:
“Forget the proverbial chicken or the egg; today we are called to answer which came first, the research or the research credit study?”
That is not startup drama. It is a court describing the core risk behind a sloppy R&D credit process. Writing a polished R&D story after the fact, without evidence that the work happened the way you say it did.
This post breaks down the case, George v. Commissioner (T.C. Memo. 2026-10), in plain English and turns it into a practical playbook for founders who want the credit and want to sleep at night.
Quick Refresher: What the R&D Tax Credit Is for Founders
The federal R&D credit is based on Qualified Research Expenses (QREs). These generally include wages, supplies, certain contract research costs, and in some cases computer use costs. All of those costs must be tied to activities that meet the legal definition of qualified research.
If you are an eligible Qualified Small Business, you may be able to elect to use up to $500,000 of the credit against payroll taxes. This is especially useful when you are not profitable yet. However, that election must be made on a timely filed return. You generally cannot go back and make the payroll election on an amended return.
Yes, the credit can be meaningful.
But it is also documentation heavy by design. The IRS makes clear that you must retain records in sufficiently usable form and detail to support what you claimed.
What Happened in George v. Commissioner
The taxpayer in this case was not a tech startup. It was a poultry business claiming R&D credits for seven “research trials” intended to create an improved poultry product.
The IRS disallowed the credits and argued that the trials were essentially a retroactive rewrite of routine operations to fit the Section 41 framework.
The court did not adopt an extreme position. Instead, it divided the projects into two categories:
- Trials supported by credible testimony and contemporaneous business records
- Trials that could not be substantiated or where records contradicted the narrative
In the end, the court allowed significant Qualified Research Expenses for certain years. However, it reduced the effective credit rate because earlier base years were not substantiated.
Founder translation: Even when you win, weak documentation can still cost you real money.
The Four Part Reality Check: What “Qualified Research” Looked Like to the Court
The opinion walks through the legal test in detail. At founder altitude, here is what matters.
To qualify, you generally need to show the work:
- Focused on improving a specific business component such as a product or production process
- Was technological in nature
- Aimed to resolve real uncertainty about capability, method, or design
- Involved a genuine process of experimentation
You will notice something. That is how strong product teams already build.
The gap is usually not behavior. It is documentation.
Where the Taxpayer Lost: When Records Do Not Match the Story
Several losses in this case are painfully relatable to startups.
In one trial, the court denied the credit because contemporaneous feed recipe records did not support the claimed experimental changes. The court refused to rely on testimony when the company’s own records told a different story.
In another trial, the records showed dosages remained constant. That directly undercut the claim that the team was experimenting with varying levels to identify an optimal approach.
In a vaccine related trial, witnesses could not establish whether the work occurred in 2012 or 2013. Without a reliable timeline, the court could not apply the year by year credit rules.
In yet another trial, the record did not clearly identify which experimental groups received which treatment. Conflicting timelines and weak substantiation led the court to deny the credit.
Founder lesson: Your claim usually does not fail because you did not innovate. It fails because you cannot prove what happened in a way that aligns with your own records.
Where the Taxpayer Won: This Looks Like Real Product Development
There were also clear wins.
One genetic line trial stood out. The court described it as the cleanest example of the scientific method in the record. There was a hypothesis, control groups, measurable outcomes, and documented analysis. The related costs were allowed.
Some probiotic work also qualified. The company did not simply implement a change and hope for better results. It ran structured trials, collected data, and analyzed outcomes.
However, even within that category, the court drew careful lines. One probiotic was disallowed because the company could not clearly identify the experimental groups outside of the credit study itself.
Founder lesson: Qualification is often not the issue. Traceability is.
The Most Founder Relevant Technical Concept: Prototypes and Pilot Models
The opinion discusses the idea of a “pilot model” under Section 174. A pilot model is something produced to evaluate and resolve uncertainty during development. Costs to produce that model can qualify, even if the product is later sold.
This is where the poultry case becomes directly relevant to startups.
If your company builds prototypes, beta hardware, experimental builds, pre production units, or internal versions to validate uncertainty, the conceptual framework is similar. The costs of building and testing those versions may fall within the development framework.
Founder translation: Your prototype work may qualify. But only if you can clearly show the uncertainty you were resolving and how the build was part of a structured experiment.
A Documentation Playbook That Will Not Slow Your Team Down
Founders often hear “document everything” and tune out. That is not realistic.
The goal is much simpler. Maintain enough evidence that a skeptical outsider can connect:
- What you were trying to improve
- What you did not know at the start
- What experiments you ran to resolve that uncertainty
- What costs map to those experiments
A lightweight approach that aligns with how strong teams already operate:
- Create a one page Experiment Brief per project.
Document the business component, the technical uncertainty, your hypothesis, and what success or failure would look like. - Tie systems of record to that brief.
For software teams, this could include tickets, pull requests, design docs, and test results. For hardware or biotech teams, this could include build notes, protocols, lab notebooks, and QA data. - Track who did what to support wage allocations.
Wages are a core QRE category. Even without traditional timesheets, you can often build a defensible approach using role based allocations supported by real artifacts. - Map expenses to experiments with a clear ledger trail.
In George, the court repeatedly asked which experimental group received what and when. Where the taxpayer could not answer that question with records, the credit failed.
Your R&D credit partner matters more than you think
There's a fascinating subplot in this case around penalties. The IRS didn't just challenge the credits — they sought accuracy-related penalties under Section 6662(a). If the court had agreed, George's would have owed additional penalties on top of any denied credits.
But the court found that George's had "reasonably relied in good faith" on the advice of the firm that conducted their study. The thoroughness of the study — nearly 100 pages, based on multiple on-site visits, extensive employee interviews, and comprehensive data analysis — demonstrated that the taxpayer took the process seriously.
This cuts both ways. It means working with a credible, thorough partner can protect you even if some claims don't survive scrutiny. But it also means that a sloppy or superficial credit study could leave you exposed — not just to denied credits, but to penalties.
For startup founders, the takeaway is simple: the firm you choose to prepare your R&D credit study isn't just a vendor. They're your insurance policy. The quality of their work directly determines your risk if the IRS ever asks questions.
At Staxiom, we take this seriously. Our co-founder spent 13 years running the West Coast R&D Tax Credit practice at Ernst & Young. We combine that Big Four rigor with automation that makes the process faster and more accurate — because we believe you shouldn't have to choose between thoroughness and efficiency.
What this means for you as a founder
Let's bring this home. If you're an early-stage startup, here's your practical checklist based on George v. Commissioner:
- Claim the credit if you're doing qualifying work. If a chicken farm can qualify, you almost certainly can. Early-stage startups without revenue can apply federal R&D credits against payroll taxes — up to $500,000 per year. That's real money back in your pocket.
- Document as you go, not after the fact. The single biggest risk factor in this case wasn't whether the research was real — it was whether there was contemporaneous proof. Keep records of what you're building, what technical challenges you're solving, and how you're evaluating different approaches. Your future self (and your future credit study) will thank you.
- Choose your R&D partner carefully. A thorough, well-documented credit study protected George's from penalties even on claims the court denied. A rushed or superficial study could leave you exposed. Look for a partner who understands the technical substance of your work, not just the tax code.
- Don't wait. George's filed amended returns years after the fact. They still got credits — but the retrospective approach created unnecessary risk and complexity. The earlier you start building your R&D credit documentation into your workflow, the stronger your position.
Ready to get your R&D credit right?
We built Staxiom to make R&D tax credits accessible, accurate, and defensible for startups. We combine deep tax expertise with automation so you get every dollar you're entitled to — backed by documentation that holds up to scrutiny.
