Most businesses claiming the R&D tax credit spend their energy on the actual work: building products, improving processes, and solving hard technical problems. The tax paperwork tends to become a year-end scramble, pulled together only when it's time to file. That's understandable. It's also exactly where the trouble starts.

The credit itself is generous, and it's completely legitimate. But it has also become a bigger target for IRS examination, especially for startups and smaller companies that filed quickly or used low-cost software to generate a claim. The good news is that the issues we see almost never involve bad intent. They're the same avoidable mistakes, over and over.

What are the most common R&D tax credit mistakes that trigger IRS problems?

The most common R&D tax credit mistakes are weak or reconstructed documentation, claiming work that fails the four-part test, including funded research, allocating wages by job title, mishandling Section 174, leaving Form 6765 Section G incomplete, skipping state credits, and claiming without a defensible strategy.

Here's a closer look at each one and how to keep your claim clean.

Mistake 1: Recreating Your Documentation After the Fact

This is the big one. When a business waits until tax season to assemble its R&D records, it ends up reconstructing the year from memory: estimating who worked on what, guessing at hours, and writing project descriptions long after the work wrapped up.

The IRS expects contemporaneous documentation, meaning records created while the research is actually happening. Project notes, design iterations, test results, and real-time tracking carry weight. A spreadsheet you built the week before filing does not. When an examiner can't connect a specific expense to a specific activity, that expense is the first thing to get knocked out.

Picture a software company that claims a chunk of its engineering payroll but rebuilds the hours from memory after year-end. On exam, there's nothing tying those wages to qualifying work, so the wage portion of the claim gets reduced or thrown out entirely.

A better approach is to capture documentation throughout the year instead of trying to recreate it at tax time. Tie your records to specific business components and qualified expenses so the story writes itself. (For what actually counts as a qualified expense, our small-business guide breaks it down.)

Mistake 2: Claiming Work That Fails the Four-Part Test

Not everything that feels innovative qualifies. To count as research, an activity must satisfy all four parts of the IRS test: it must have a permitted purpose (improving a product or process), aim to eliminate technological uncertainty, involve a process of experimentation, and be technological in nature.

The word that matters there is "all". Miss one part and the activity doesn't qualify, no matter how much effort went into it. Routine engineering, cosmetic tweaks, market research, and ordinary quality control tend to fail, even though they often appear in claims.

Say a manufacturer claims the time its team spent adjusting existing tooling or giving a product a cosmetic refresh. That's normal business work, not experimentation, and it's the kind of thing an examiner flags quickly.

Before you claim an activity, run it through all four parts and ask the question the IRS will: could this survive a process-of-experimentation challenge? If you're unsure whether an activity qualifies, pause before you include it. Spending a little more time on it now is far easier than explaining it during an IRS exam. If you're unsure whether an activity qualifies, our federal R&D tax credit specialists can help you make the call before you claim it rather than defending it later. (Our explainer covers what qualifies for the R&D credit in more depth.)

Mistake 3: Claiming Funded or Customer-Paid Research

Here's one that catches a lot of contract shops, agencies, and product-development firms off guard. If someone else pays for the research and takes on the risk, it usually isn't yours to claim.

The rule comes down to two questions: who bears the financial risk, and who keeps the rights to the results? If a customer pays you a fixed fee regardless of whether the work succeeds and they walk away owning the output, that's "funded research," and it's excluded. You can't claim credit for research you didn't actually fund.

Think of a development firm building software for a client under contract. The client pays no matter what, and the client owns the finished product. The firm never assumed the financial risk, so those costs don't qualify, even though real engineering work was done.

This is one of the most-missed exclusions and one of the most common adjustments on the exam. If your work involves client contracts or grants, look closely at the risk-and-rights question before you include anything.

Mistake 4: Allocating Wages by Job Title Instead of Time

Wages are usually the largest component of an R&D claim and one of the main categories of qualified research expenses (QREs), making them the component the IRS scrutinizes most. One of the biggest wage allocation errors is relying on job titles or department averages instead of documenting what employees actually spent their time on.

Claiming 100% of an "engineering department" (managers, admins, and everyone), or applying a flat 75% across the board, doesn't hold up. The credit requires a real connection, the technical term is "nexus", between the wage and the qualified activity. There's also a "substantially all" rule that lets you count an employee's full wages when at least 80% of their time is spent on qualified work, but that's a threshold to be supported by records, not a shortcut to apply to everyone.

The cleaner approach is to base allocations on actual time, through contemporaneous tracking or documented interviews, so each dollar you claim ties back to qualifying work. If you're a startup using the credit against payroll taxes, getting these allocations right matters even more. Our guide for startups walks through how that offset works.

Mistake 5: Mishandling the Section 174 Capitalization Rule

This is the area where the rules have moved and where we see the most confusion right now. As of 2026, the treatment works as follows: domestic research costs are immediately deductible under the newer Section 174A rules, while foreign research generally must be capitalized and amortized over 15 years.

This is where many businesses get confused because two different tax rules are involved. Section 174 governs how you treat research expenses for tax purposes. The Section 41 research credit is a separate calculation that sits on top, and there's an interaction with Section 280C that affects how much of your deduction you keep when you claim the credit. The two have to line up.

Many businesses run into trouble by treating these rules too casually: deducting everything without separating domestic from foreign research, ignoring the required amortization on foreign work, or mishandling the Section 280C election. Each one creates an inconsistency between your deduction and your credit, and inconsistencies are exactly what examiners look for.

Mistake 6: Leaving Form 6765 (Section G) Incomplete

The IRS redesigned Form 6765, and the new Section G asks for business-component-level detail it never required before: a breakdown of qualified expenses by component, plus a description of the information each project was trying to discover. For 2026 returns, that detail matters more than it used to.

When that section is vague or left half-finished, it shows. The IRS now reviews claims more carefully from the start. If Section G lacks clear business-component detail, your claim can draw extra scrutiny before anyone even looks at the underlying research. You did the work, but the form didn't tell the story.

Completing Section G correctly isn't complicated, but it does take care. Fill it out with specific, project-level detail that matches your documentation. You can find the exact fields in the IRS Form 6765 instructions, and since the requirements changed for 2026, it's worth understanding exactly what changed on Form 6765 before you fill it out.

Mistake 7: Overlooking State-Level R&D Credits

The federal credit gets all the attention, but it's often only part of the picture. Most states with an income tax offer their own R&D credit, and in many cases, it stacks right on top of the federal one. Claiming federal only leaves real money unclaimed and can create a federal/state mismatch that warrants a second look.

Because CMP works with businesses across the country, we check this in every engagement. The details vary a lot by state. Utah, California, and Georgia all have research credits, but the rates, definitions, and carry-forward rules aren't the same. What qualifies federally usually qualifies at the state level too, but you have to file for it and coordinate the two claims so they tell a consistent story.

If you operate in Utah, there's a Utah R&D tax credit that pairs with the federal credit, and it's a common one to miss. Wherever you're based, the question is the same: are you claiming every credit your research earned, not just the federal one?

Mistake 8: Claiming Without an Audit-Ready Strategy

The last mistake is about mindset more than any single line item. R&D claims tend to fail at the two extremes.

On one end are the aggressive promoters: firms that promise an eye-popping credit, claim everything in sight, and leave you holding the bag if the IRS pushes back. On the other end are business owners who are so worried about an audit that they under-claim or skip the credit entirely, leaving money behind. Neither is a strategy.

Overstated claims can trigger an exam, a disallowance, and accuracy-related penalties of up to 20% under the IRC §6662 rules. Being too cautious just costs you the benefit you legitimately earned. The goal is the middle: a claim that's as large as the facts support and documented well enough to defend if anyone asks.

That balance is exactly what a CPA who does this work regularly brings to the table: confidently claiming and keeping it defensible.

How the Eight Mistakes Compare at a Glance

Frequently Asked Questions

What triggers an IRS audit of the R&D tax credit?

The most common triggers are weak or after-the-fact documentation, claims that include routine or non-qualifying work, large credits relative to company size, and incomplete Form 6765 reporting. Refund claims and startup payroll-offset claims have drawn extra scrutiny in recent years.

What documentation do I need to defend an R&D claim?

You want contemporaneous records: project notes, design iterations, test results, and time tracking created while the work happened. They should tie specific expenses to specific business components. Reconstructed or estimated records are far weaker, and they're usually the first thing reduced on exam.

What activities fail the four-part test?

Activities fail when they lack any of the four parts: a permitted purpose, technological uncertainty, a process of experimentation, and a technological nature. Routine engineering, cosmetic changes, market research, and ordinary quality control typically don't qualify, even when they feel innovative.

How does Section 174 affect my R&D claim?

Section 174 controls how you treat research costs, separate from the Section 41 credit. As of 2026, domestic research is generally deductible immediately under Section 174A, while foreign research is amortized over 15 years. The two have to reconcile, so confirm the current treatment with your CPA before filing.

Can I claim a state R&D credit in addition to the federal credit?

Usually, yes. Most states with an income tax offer their own R&D credit that stacks with the federal one, though rates and rules vary by state. Claiming federal only leaves state money unclaimed and can create a federal/state mismatch.

What happens if the IRS disallows my R&D credit?

A disallowed credit can mean repaying the benefit, plus interest and potential accuracy-related penalties of up to 20% under IRC §6662. In some cases, the IRS may examine related years as well. Strong documentation and a defensible methodology are your best protection if a claim is challenged.

How far back can I amend to claim, or be audited on, the credit?

You can generally amend to claim the credit for the past three tax years, subject to specific rules, and the IRS generally has three years to examine a return (longer in certain situations). Amended refund claims also require extra detail, so they're worth preparing carefully.

Keeping Your R&D Claim Audit-Ready

Every one of these mistakes comes back to the same two habits: keep good records while the work is happening, and have someone experienced review the claim before you file. The credit is worth real money, and you've earned it by doing the work. A little discipline in how you capture and report it is what keeps that money in your pocket if the IRS ever asks questions.

At CMP, our federal R&D tax credit services team works with businesses across the country to claim the credit they're entitled to, and to do it in a way that holds up. If you want to talk through how these issues apply to your situation, or you'd like a second look at a claim before you file, we're happy to help.