Federal research tax credit litigation illustrates the importance of proving both the nature of research activities and the expenses attributable to them. Section 41 of the Internal Revenue Code provides the credit for increasing research activities, commonly called the research and development (R&D) tax credit. Courts apply its statutory tests to the evidence for the relevant business components. Coleman v. Commissioner, T.C. Memo. 2020-146, concerns gambling-loss substantiation, not R&D credits; it should not be identified as Redelsheimer v. Commissioner or presented as a research-credit holding.

Cases rejecting unsupported allocations and generalized descriptions of engineering work demonstrate why taxpayers should connect technical uncertainty, experimentation, and claimed costs. They do not establish a universal prohibition on retrospective evidence. This study examines the statutory framework, evidentiary issues, and practical implications of the cases discussed below.

The Statutory Architecture of the Research and Development Tax Credit

The R&D tax credit encourages qualifying domestic research, but an expenditure’s treatment as research expenditure does not by itself establish credit eligibility. The cases discussed here generally concern earlier tax years governed by Section 41’s former cross-reference to Section 174. Following Public Law 119-21, Section 41(d)(1)(A) refers to domestic research or experimental expenditures under Section 174A, generally for amounts paid or incurred in taxable years beginning after December 31, 2024. Section 174A permits current deduction of qualifying domestic expenditures, subject to its rules and elections; foreign research remains outside the credit. Historical cases must be read under the law applicable to their tax years.

Qualified research expenses (QREs) include eligible wages, supplies, certain computer-use costs, and the allowable portion of contract research expenses. Contract research expenses generally enter the calculation at 65%, subject to statutory exceptions. The regular credit generally uses 20% of the excess of current-year QREs over the base amount. The alternative simplified credit generally uses 14% of the excess over 50% of average QREs for the three preceding tax years, with a separate rule when any such year has no QREs. Base-amount limitations, elections, and other statutory adjustments must also be considered.

Section 41 defines qualified research through cumulative requirements commonly described as the four-part test. Eligibility must also be assessed against the statutory exclusions.

The Four-Part Test for Qualified Research

The tests initially apply separately to each business component: a product, process, computer software, technique, formula, or invention held for sale, lease, or license, or used in the taxpayer’s trade or business. Failure at the overall component level does not necessarily eliminate every associated expense. Treasury Regulation Section 1.41-4(b)(2) provides a shrinking-back rule for examining the most significant qualifying subset of elements.

Substantiation Standards: Cohan and Reliable Evidence

Taxpayers generally bear the burden of proving entitlement to a claimed credit, subject to applicable burden-shifting rules. The Cohan doctrine permits estimation in appropriate circumstances when deductible spending has been established and there is a reasonable evidentiary basis for an estimate. In Coleman v. Commissioner, T.C. Memo. 2020-146, the Tax Court considered financial records, testimony, and statistical evidence in determining whether gambling losses equaled or exceeded winnings for 2014. Expert Mark Nicely estimated that the chance of a net gambling profit was approximately one in 140 million under his analysis.

Coleman illustrates the potential value of corroborated expert evidence, but it does not resolve the distinct requirements of Section 41. Establishing that money was spent is different from establishing that the underlying activities were qualified research. An estimate must rest on evidence adequate for the particular statutory issue.

The Limits of Expert Testimony in R&D Claims

The statistical evidence in Coleman was tied to the taxpayer’s gambling activity and supported by other financial evidence. Its reasoning cannot simply be transferred to engineering projects. In an R&D claim, expert analysis should explain the actual work performed, the uncertainties addressed, and the basis for attributing expenses to that work.

Siemer Milling Co. v. Commissioner, T.C. Memo. 2019-37, and Little Sandy Coal Co. v. Commissioner illustrate the consequences of failing to demonstrate experimentation through the evidence presented. Neither creates a universal rule that only contemporaneous time sheets can establish a credit. The persuasive value of testimony, reconstructed allocations, and technical records depends on their specificity and reliability.

The “Substantially All” Rule and the Shipbuilding Precedent

The “Process of Experimentation” (PoE) test requires that “substantially all” of the research activities must constitute elements of a process of experimentation. In this context, “substantially all” is defined as 80% or more, measured by cost or another consistently applied reasonable basis.

Analysis of Little Sandy Coal Co. v. Commissioner

In Little Sandy Coal Co. v. Commissioner, 62 F.4th 287 (7th Cir. 2023), the Seventh Circuit affirmed the disallowance of the shipbuilder’s credit while disagreeing with parts of the Tax Court’s reasoning. The litigation examined a tanker and dry dock as representative vessels. Key points included:

Business Components and Shrinking Back: An entire vessel may be an appropriate initial business component. Where the whole component fails, shrinking back requires evidence supporting a qualifying subset; identifying a large component was not itself the legal error.

Production Activities: Producing a pilot model can constitute an element of experimentation. The appellate court rejected categorical exclusion of direct support and supervision from the numerator. The taxpayer still had to establish which activities qualified.

Employee-Time Allocations: The taxpayer did not provide a principled way to separate experimentation-related activities from other work. Arbitrary estimates and the novelty of a vessel did not establish the required percentage.

For the tax years considered in Little Sandy Coal, the substantially-all comparison was the cost, or another consistently applied reasonable measure, of research activities constituting elements of experimentation divided by all relevant research activities for the business component. The required result is at least 80%. This is not a percentage of the component’s physical newness or an automatic comparison with every production cost.

Direct supervision, direct support, and pilot-model production must be evaluated according to their role in the research. Routine activity does not become experimentation merely because it occurs on an innovative project. Conversely, the job title of a production worker does not automatically exclude otherwise qualifying work.

Technological Uncertainty and the Discovery of Information

Technical uncertainty exists when the information available to the taxpayer does not establish the capability or method for developing or improving the business component, or its appropriate design. The uncertainty must be connected to the development effort. Commercial uncertainty, such as whether customers will buy a product, does not by itself satisfy the technological requirements.

Engineering Calculations vs. Experimentation

In Phoenix Design Group, Inc. v. Commissioner, T.C. Memo. 2024-113, the Tax Court examined three sample projects undertaken by a mechanical, electrical, plumbing, and fire protection engineering firm. It found that none demonstrated qualified research, including at the shrinking-back level. Applying established calculations to available information did not, on the evidence presented, establish the claimed investigative process. The decision does not categorically exclude engineering calculations, iterative design, or engineering firms from the credit.

Existing scientific principles can be used in qualified research. Treasury Regulation Section 1.41-4(a)(3) does not require discovery beyond the common knowledge of skilled professionals in the field. Technical literature may help explain what the taxpayer knew, but it is not a universal industry-novelty benchmark. The source’s suggested connection between milling publications by Elena Redelsheimer and the cited tax holdings does not establish a legal test and should not be used to determine eligibility.

The Adaptation Exclusion

Section 41(d)(4)(B) excludes adaptation of an existing business component to a particular customer’s requirement or need. Custom work is not automatically excluded: Treasury Regulation Section 1.41-4(c)(3) states that the exclusion does not apply merely because a component is intended for a specific customer. In Betz v. Commissioner, T.C. Memo. 2023-84, the court rejected claims involving air-pollution-control systems on the record presented, including failures to establish qualifying uncertainty and experimentation and funded-research issues. Testing compliance with customer specifications does not, by itself, establish uncertainty at the outset.

Funded Research and the Allocation of Financial Risk

Under Section 41(d)(4)(H), research is excluded to the extent funded by another person. Treasury Regulation Section 1.41-4A(d), incorporated by Section 1.41-4(c)(9), requires analysis of the agreements and the parties’ rights and obligations. Two central issues are:

Substantial Rights: If the performer retains no substantial rights in the research results, the research is treated as fully funded. Exclusive ownership is not required, but incidental experience alone is insufficient.

Economic Risk: Payments contingent on successful research are not treated as funding in the same manner as payments owed regardless of outcome. The extent of funding depends on the arrangements and applicable rules; a contract label alone is not decisive.

Contractual Analysis in the 2025 Proceedings

The early-2025 proceedings in Smith v. Commissioner and System Technologies, Inc. v. Commissioner addressed IRS requests for summary judgment on funded research. Denial of such a motion is not a final award of every claimed credit. Their procedural posture is essential to interpreting the outcomes.

Design Milestones: In Smith, the taxpayers argued that entitlement to payment depended on successful completion of design milestones. The court found unresolved issues precluding summary judgment; this was not a categorical holding that milestone billing proves eligibility.

Governing Law: Smith involved issues concerning foreign governing law. In System Technologies, Indiana law and available remedies for failure to deliver affected the court’s assessment of whether payment depended on success.

Retained Rights: Contractual rights to use research results must be examined separately from payment risk. Sharing rights or providing design documents to a customer does not automatically resolve the substantial-rights question.

These proceedings illustrate why contractual language, applicable law, and procedural posture matter. They should not be characterized as eliminating contractual analysis or guaranteeing credits to contractors.

Documentation and the “Usable Form” Requirement

Documentation is a frequent issue in research-credit disputes, but claims can also fail because the activities or expenditures do not satisfy substantive eligibility rules. Treasury Regulation Section 1.41-4(d) requires records adequate to substantiate the claimed expenditures. It does not prescribe a single mandatory record format.

The Role of Contemporaneous Records

Records created during a project are often persuasive because they show what was uncertain and what was actually evaluated. Time allocations should be supported by a reasoned method and linked to relevant activities. Design revisions, correspondence, test results, and credible testimony can contribute to the evidentiary record. There is no universal requirement that every claimant maintain a particular testing log or time-tracking system.

Implications for Future R&D Applications

The R&D cases discussed here support three practical areas of attention: identifying business components, substantiating technical activities and costs, and examining contractual rights and risk. Coleman offers only a general evidentiary analogy; it is not an R&D-credit decision.

Pillars of Modern R&D Compliance

The first area is project granularity. Begin with the appropriate business component and apply shrinking back where warranted. Taxpayers are not required to start at the lowest conceivable component level, but they need adequate evidence for any subset relied upon.

The second area is technical documentation. Engineering and accounting records should work together: technical records explain uncertainty and evaluation, while financial records support the expenses claimed. Technical publications can provide context, but neither publication nor advancement beyond industry knowledge is a condition of eligibility.

The third area is contractual alignment. Review payment obligations, acceptance terms, remedies, and rights to use research results under the governing law. The analysis must reflect the actual arrangements. Fixed-price terminology alone does not establish qualifying financial risk, and contractual revisions do not retroactively alter the facts of completed research.

Technical Nuances of the Process of Experimentation

Systematic trial and error can constitute a process of experimentation. Treasury Regulation Section 1.41-4(a)(5) identifies modeling, simulation, and systematic trial and error as possible evaluative methods. Unstructured attempts, routine adjustments, or assertions that a project was innovative do not by themselves establish the test. The taxpayer must explain the uncertainty, the alternatives, and how they were evaluated.

Modeling and Simulation as PoE

Siemer Milling illustrates the need to explain the evaluation process rather than simply describe product development. Modeling and simulation can be relevant evidence, but neither is mandatory in every case. For example, a reliability model used to compare uncertain bearing designs or an optimization model used to evaluate uncertain manufacturing alternatives may support a claim. These are illustrative possibilities, not automatic qualifications or identified judicial holdings about those particular methods.

Where models are used, retain their assumptions, inputs, outputs, revisions, and the resulting engineering decisions. A sophisticated mathematical model does not alone prove qualification: it must be connected to the taxpayer’s uncertainty, evaluation of alternatives, relevant business component, and claimed expenses.

Final Thoughts

Section 41 requires proof that the claimed activities and expenditures satisfy the applicable statutory and regulatory requirements. Little Sandy Coal, Betz, and Phoenix Design Group demonstrate the limits of relying on project novelty or generalized engineering descriptions. Coleman remains a separate gambling-loss decision and should not be treated as an R&D precedent.

For future claims, a practical approach is to document technical questions and evaluations while the work is underway, link expenses to the activities performed, and analyze relevant contractual arrangements. Credible narratives and reasonable retrospective estimates may supplement the evidence; their value depends on the factual foundation. No single document type guarantees a credit.

Businesses should apply the law for the relevant tax year, including the Section 174A changes where applicable. The governing benchmark is the statutory definition of qualified research, supported by reliable evidence, rather than the novelty of a product, the sophistication of a mathematical technique, or comparison with a particular academic publication.