Claiming the R&D Credit Was the Easy Part: What Defending It Actually Takes

1 — Computing the Credit Is Only Step 13 of 16

A research tax credit study runs through a long sequence before a dollar figure is final: assess the need for a study, define the engagement, identify the businesses, years, and sources involved, identify and exclude qualified versus nonqualified research and expenses, identify gross receipts, make allocations, and finally compute the credit. Two more steps follow before the study is done — considering alternatives and preparing deliverables — and defending the credit is the last step of all: not a separate, optional add-on, but the stage the entire study has been building toward.

2 — Why the IRS Started Scrutinizing These Claims Harder

The IRS's own Audit Techniques Guides flagged a specific pattern: taxpayers submitting "prepackaged" research credit claims, often delivered in binders by representatives working on a contingent-fee basis, that examiners frequently found did not substantiate the actual qualified research expenses claimed. Then-IRS Commissioner Mark Everson told the Senate Finance Committee in 2006 that many of these refund claims relied on unsupportable amounts, nonqualified expenditures, or estimates lacking contemporaneous documentation — and that late-filed claims, often arriving well into the audit cycle, placed a heavy burden on examination teams. The guidance is explicit that examiners should not rely solely on critiquing a taxpayer's prepackaged methodology; they're expected to independently determine what documentation and testimony is actually needed.

3 — What an Audit Team Looks Like, and the Three Fights That Recur

A research credit audit can involve a full team: the IRS auditor, engineer agents and their managers, technical advisors, and sometimes a computer audit technician, with an IRS attorney available for legal support. The audit is built around Information Document Requests (IDRs) — a standardized initial questionnaire followed by others shaped by the taxpayer's responses — and the IRS is required to bring in outside software experts for any internal-use-software credit averaging $500,000 or more annually. Across all of this, the same three issues come up again and again: whether the research itself is "qualified" (tested under a four-part scientific-manner standard), whether the taxpayer's substantiation is sufficient, and whether the dollar amounts were correctly determined.

4 — Preparing Before the IRS Even Sends the First Letter

Once the initial audit letter arrives, the recommended response isn't reactive — it's a structured five-part preparation: anticipate the issues an auditor is likely to raise (often signaled by published IRS guidance and coordinated issue papers), marshal the evidence tied to specific business components and research activities, review the credit computation itself for technical errors (wrong base-period years, misapplied exclusions, incorrect gross receipts), stress-test the computation by rerunning it with larger and smaller QRE and gross-receipts figures to see which variables actually move the credit, and look for additional QREs or gross receipts that were left out — including contract expenses that were actually misclassified wage, supply, or computer-rental expenses, which is a common and often-overlooked way to increase the credit during defense rather than shrink it.

5 — Managing the Audit's Scope and Its Clock

Taxpayers have real tools to control how big and how long an audit gets. A Limited Issue Focused Exam (LIFE audit) is a streamlined, agreed-scope process that can keep the audit from sprawling past the research credit into other issues. Separately, the audit's duration is bounded by the Assessment Statute Expiration Date (ASED) — generally three years from filing — which the IRS typically wants extended using a Form 872 (which extends the ASED to a specific date and limits consent to named issues) or a Form 872-A (an indefinite extension the taxpayer can terminate). Taxpayers are not required to agree to extend the ASED at all, though refusing will generally trigger a formal 90-day deficiency notice.

6 — If the Audit Doesn't End in Agreement: The Appeals Path

When the IRS and the auditor can't resolve the credit, the case moves toward a 30-day letter (with appeal rights), a 90-day deficiency notice, or a claim disallowance letter — any of which can route to the IRS Appeals Office, which handles roughly 30,000 to 50,000 cases a year with about 1,000 appeals officers. Appeals is explicitly built around settling cases based on the perceived litigation hazards for both sides, not on reaching the single "technically correct" answer — and appeals officers are barred from communicating with the original auditor about the issues without giving the taxpayer a chance to participate, to protect the independence of the appeals process. For certain research-credit issues designated as an "Appeals Coordinated Issue," the appeals officer must also consult with and get concurrence from a technical guidance coordinator before finalizing a settlement, which limits how much authority the local appeals officer actually has on his own.

7 — Three Courts, One Choice, Three Different Tradeoffs

If a credit dispute isn't resolved at audit or appeals, the taxpayer can choose to litigate in one of three forums, each with real structural differences. The U.S. Tax Court is the only one of the three that lets a taxpayer litigate without first paying the disputed tax — a significant practical advantage — but it's a national court based in Washington, D.C. with no jury and limited to tax cases specifically. The U.S. Court of Federal Claims requires the taxpayer to pay the tax first and sue for a refund, also has no jury, but can hear a broader range of claims against the government (patent, customs, contract, and tax). Federal district courts are courts of general jurisdiction — they allow a jury, don't specialize in tax, and apply the full federal rules of evidence and local procedural rules, making the litigation more like ordinary civil litigation than a specialized tax proceeding.

8 — What Actually Happened in the Cases Courts Have Decided

The reported cases show these principles playing out concretely. In United States v. McFerrin, the government sued to recover an allegedly erroneous refund; the district court sided with the government and even declined to apply the Cohan doctrine (which lets a court estimate expenses when exact records are incomplete) — but the Fifth Circuit reversed and sent the case back specifically to apply Cohan and estimate the taxpayer's research expenses. In FedEx Corp. v. United States, the dispute turned on whether the taxpayer could rely on older regulations' internal-use-software provisions that the IRS had simply left out of newer regulations; the court held the taxpayer could rely on either set of rules, rejecting the government's position that it could refuse to issue regulations and then penalize taxpayers for the resulting gap. In Trinity Industries, a boat manufacturer lost because it applied an all-or-nothing 80%-rule shortcut to identify qualifying expenses rather than examining its research records project by project, in contrast to the far more granular approach the Tax Court took in Union Carbide's case. And in TG Missouri, the Tax Court sided with the taxpayer on whether third-party mold-manufacturing costs counted as qualifying "supply" expenses. The government can also go on offense: in McFerrin, it sued the taxpayer directly to recover a refund it believed was issued in error — something it's generally entitled to do within two years of the erroneous refund, extendable by agreement.

The big picture

Defending a research tax credit isn't a single event — it's a sequence of distinct legal forums, each with its own rules, incentives, and leverage points. Only the last is a court. The sequence runs from the audit itself to the Appeals Office and then, if it gets that far, to the U.S. Tax Court, the Court of Federal Claims, or a federal district court. The credit's survival depends less on the size of the number claimed and more on whether the underlying research, expenses, and methodology can be reconstructed and defended at every one of those stages, often years after the research actually happened.

In your own words: why did Trinity Industries lose using the same 80% "substantially all" rule that auditors themselves often apply, while the Tax Court took a much more granular, project-by-project approach in the Union Carbide case — and what does that difference suggest about which kind of documentation actually survives court scrutiny?

You just read about it. Could you teach it?

Reading a good explanation feels like understanding it. Usually it isn't the same thing — and you don't find out which one you've got until someone asks you to explain it back.