The credit is not limited to laboratories — software development, process improvement and product engineering can qualify. It turns on a four-part test and on documentation tying wages and hours to specific technical uncertainty, kept as the work happens rather than reconstructed by a consultant afterwards.
Two things are true at once. The credit is genuinely broader than most owners assume, and the marketing around it substantially overstates who qualifies.
What the test asks
Broadly: is the work technological in nature, is it aimed at a permitted purpose such as improving a product or process, was there genuine uncertainty at the outset, and did you resolve it through a process of experimentation.
All four. The one that fails most often is uncertainty — implementing a known solution competently is not research, however difficult it was.
Activities that surprise people
Developing software. Designing a manufacturing process. Engineering a product to meet a new specification. Formulating something. None of it requires a laboratory or a person with the word research in their title.
What it requires is that you did not know at the start whether the approach would work, and that you found out by trying.
Activities that do not qualify
Routine configuration of purchased software. Cosmetic changes. Market research. Quality control on an established process. Adapting an existing product for a particular customer without technical uncertainty.
Also anything done after commercial production begins, and generally anything funded by a customer who bears the risk.
Documentation is the whole battle
The credit is claimed on wages, supplies and contract research. Wages dominate, which means the evidence is time — who spent how long on which qualifying project.
Businesses that track time to projects already have most of it. Businesses that do not are asking people to reconstruct a year from memory, which produces a study that reads as reconstructed.
The contingency-fee firms
They are not all bad and the model is not inherently a problem. What matters is whether the study is defensible on your facts.
Ask what happens if the position is examined, who supports it, and whether the fee is affected. You sign the return; the consultant does not.
The payroll offset
Qualifying small businesses can apply part of the credit against payroll tax rather than income tax. For a pre-profit company that is the difference between a credit that helps now and one that waits for profitability.
There are eligibility conditions and an election to make on time, so it is a conversation to have before filing rather than after.
Where to start
List the projects where you genuinely did not know whether something would work. If that list is empty, the credit is probably not yours. If it is not empty, your CPA can tell you quickly whether it is worth a study.
Common questions
- Do I qualify for the R&D tax credit?
- Possibly, if you are resolving genuine technical uncertainty through a process of experimentation. It is a four-part test your CPA applies to your actual activities.
- Does software development count?
- It can. Internal-use software faces additional requirements, and routine configuration or standard implementation generally does not qualify.
- What records do I need?
- Time or project records tying named people to qualifying activity, plus evidence of the uncertainty and the experimentation. Contemporaneous beats reconstructed by a wide margin.
- What about the firms that do this on contingency?
- Some are excellent and some produce studies that do not survive scrutiny. You sign the return, so the position is yours regardless of who wrote the study.
- Can it offset payroll tax?
- Qualifying small businesses can elect to apply a portion against payroll tax rather than income tax, which matters when there is no income tax to offset.
Behind on Your Books?
We handle your bookkeeping end-to-end — categorization, reconciliation, month-end close, and clean financial statements, so your books stay current and CPA-ready.