What Activities Qualify for the R&D Tax Credit?

Last reviewed October 14, 2026

The federal R&D tax credit under IRC Section 41 rewards businesses for investing in research and development — but not every technical activity qualifies. The IRS uses a specific framework called the four-part test to determine whether an activity constitutes qualified research. Understanding this test is the first step toward identifying which of your activities generate the credit.

The IRS four-part test

Every qualifying activity must satisfy all four prongs of the test defined in IRC Section 41(d). If any one prong is not met, the activity does not qualify.

1. Business component test (IRC Section 41(d)(1)(A))

The research must relate to a business component — a product, process, technique, formula, invention, or software that the taxpayer intends to sell, lease, license, or use in its trade or business. The activity must be undertaken for a purpose related to the taxpayer's business, not purely academic or speculative research.

2. Technological uncertainty (IRC Section 41(d)(1)(B)(ii))

There must be uncertainty about the capability of developing or improving the product, the method for achieving it, or the appropriate design. This uncertainty must be technological in nature — meaning it cannot be resolved by simply consulting existing knowledge, prior experience, or readily available information. Business uncertainty (such as whether a product will sell) does not satisfy this requirement.

3. Process of experimentation (IRC Section 41(d)(1)(C))

The taxpayer must engage in a systematic process designed to evaluate one or more alternatives to resolve the technological uncertainty. This can include modeling, simulation, systematic trial and error, or other methodical approaches. The process does not need to follow the formal scientific method, but it must be more than ad hoc troubleshooting or guesswork.

4. Technological in nature (IRC Section 41(d)(1)(A))

The process of experimentation must rely on principles of engineering, computer science, biological science, or physical science. Activities grounded in social sciences, arts, humanities, or economics generally do not meet this requirement.

Qualifying activities by category

The following categories represent common areas where businesses perform work that meets the four-part test. Within each category, the critical question remains: did the work involve technological uncertainty that required experimentation to resolve?

Product development

Developing new products or meaningfully improving existing ones is one of the most common qualifying activities. This includes:

• Designing and prototyping a new product where the performance characteristics, materials, or architecture are uncertain
• Improving the reliability, speed, durability, or efficiency of an existing product through technical experimentation
• Developing new features that require solving engineering or scientific problems
• Testing alternative materials, components, or configurations to achieve specific performance targets

The improvement does not need to be revolutionary. Incremental improvements qualify as long as technological uncertainty exists and experimentation is used to resolve it.

Process improvement

Improving manufacturing processes, production methods, or operational systems through technology qualifies when it involves genuine experimentation:

• Designing new manufacturing processes to improve yield, reduce waste, or increase throughput
• Developing automation systems where the integration or performance is technically uncertain
• Experimenting with new production techniques, tooling, or equipment configurations
• Resolving technical challenges in scaling a process from prototype to production volume

Software development

Software development frequently qualifies for the R&D credit when it goes beyond routine coding:

• Building new software platforms or applications where the architecture, algorithms, or performance characteristics are uncertain
• Developing features that require solving technical problems — such as handling large-scale data processing, real-time performance, or complex integrations with external systems
• Rebuilding or re-architecting existing systems to achieve new capabilities that cannot be accomplished through routine modifications
• Developing proprietary algorithms, data structures, or system designs

Note: internal-use software faces additional requirements under the high threshold of innovation test. See the FAQ section below.

Engineering and design

Engineering activities that involve evaluating alternatives and resolving technical unknowns are strong candidates for the credit:

• Prototyping and iterative design where each iteration addresses a specific technical uncertainty
• Using CAD, finite element analysis, or computational simulation to evaluate design alternatives
• Materials testing to determine whether a material meets performance, safety, or durability requirements for a new application
• Developing custom tooling, fixtures, or dies where the design requires solving engineering problems

Scientific research

Research grounded in the physical or biological sciences qualifies when it involves systematic experimentation:

• Developing new chemical formulations or compounds with uncertain properties or performance
• Biological testing, including cell culture studies, biocompatibility testing, or microbial analysis
• Environmental testing to evaluate how materials or systems perform under specific conditions
• Pharmaceutical or agricultural research involving experimentation with new compounds, delivery mechanisms, or growing techniques

Activities that do NOT qualify

IRC Section 41(d)(4) specifically excludes certain categories of activity. Even if work appears technical, it does not generate the credit if it falls into one of these exclusions.

Market research, surveys, and social science

Consumer surveys, focus groups, market analysis, management studies, and research grounded in social sciences, arts, or humanities do not qualify. The credit applies only to research that is technological in nature.

Quality control and routine testing

Routine testing or inspection performed to verify that a product meets existing specifications is not qualified research. Testing qualifies only when it is conducted to resolve a technological uncertainty — where the outcome is genuinely unknown and informs a design or process decision.

Cosmetic or style changes

Changes related to style, taste, cosmetic appearance, or seasonal design factors are excluded under IRC Section 41(d)(3)(B), regardless of how technically rigorous the process may be. This exclusion commonly affects fashion, food and beverage, and consumer product industries.

Research after commercial production

Activities conducted after a product or process is ready for commercial sale or use generally do not qualify. Routine production monitoring, post-launch quality assurance, and minor tweaks to a finalized design fall outside the scope of the credit. However, developing a genuinely new or improved version that involves new technological uncertainty is a separate research effort and can qualify.

Funded research

Research funded by another party — through a grant, contract, or other arrangement — does not generate the credit for the party performing the research if that other party retains the rights to the research results. The critical question is who bears the economic risk: if the research fails and the funder still pays, the research is considered funded and excluded.

Foreign research

Only research performed within the United States, Puerto Rico, or U.S. possessions qualifies under IRC Section 41(d)(4)(F). Wages paid to employees performing research abroad and contractor payments for foreign research do not count as qualified research expenses.

Adaptation without technical uncertainty

Adapting an existing business component for a specific customer using established techniques does not qualify. If the customization is routine and does not involve technological uncertainty — the outcome is known and the methods are well-established — it falls under the adaptation exclusion in IRC Section 41(d)(4)(B). However, if the adaptation requires solving genuine technical problems, it may still meet the four-part test.

The “substantially all” rule

Under IRC Section 41(d)(1)(A), at least 80 percent of a taxpayer's research activities for a business component must constitute elements of a process of experimentation for a “qualified purpose” — meaning they must relate to new or improved function, performance, reliability, or quality. This is known as the substantially all test.

In practice, this means that if more than 20 percent of the activities for a given business component are directed at non-qualifying purposes (such as style or cosmetic factors), the entire business component fails the test. The substantially all rule is applied at the business component level, not at the project or company level.

Qualified research vs. routine development

The distinction between qualified research and routine development is one of the most important — and most frequently misunderstood — aspects of the R&D credit.

Qualified research involves work where the outcome is uncertain and the taxpayer must experiment to determine whether the desired result can be achieved, how to achieve it, or what the best design is. The uncertainty must be technological, not business-related.

Routine development uses well-established techniques and known methodologies where the outcome is predictable. Building a standard website using existing frameworks, writing CRUD operations for a database, or applying a known formula to a new dataset are examples of routine development that typically do not qualify.

The same project can include both. A software project might involve routine front-end layout work (non-qualifying) and complex algorithm development with uncertain performance characteristics (qualifying). Only the expenses allocable to the qualifying portion generate the credit.

Frequently asked questions

Does improving an existing product count as qualifying R&D?

Yes. Improving an existing product qualifies for the R&D tax credit as long as the improvement involves technological uncertainty and a process of experimentation. You do not need to create something entirely new. Incremental improvements to performance, reliability, or functionality can qualify if the outcome was not known in advance and you evaluated alternatives to resolve the uncertainty.

Can internal-use software qualify for the R&D tax credit?

Yes, but internal-use software must meet a higher threshold under the high threshold of innovation test set out in Treas. Reg. 1.41-4(c)(6). In addition to satisfying the standard four-part test, the software must be innovative, involve significant economic risk, and not be commercially available for purchase, lease, or license.

Do failed R&D projects qualify for the tax credit?

Yes. The R&D tax credit is based on the research process, not the outcome. If a project involved qualified research activities and incurred qualifying expenses, those expenses can generate the credit regardless of whether the project succeeded, was abandoned, or produced a commercially viable result.

What if only part of a project involves R&D activities?

You can claim the portion of expenses tied to qualifying activities. Many projects include both qualifying R&D work and non-qualifying tasks such as project management, marketing, or routine production. Only the wages, supplies, and contract research expenses allocable to the qualifying research activities generate the credit.