Most businesses that assume they don't qualify are failing a test they've misread. Here is what the IRS actually asks, without the jargon.
The most common sentence we hear on a first call is some version of "we don't do research." It is almost always said by someone whose team spent last quarter figuring out why a part kept failing at tolerance.
The disconnect is the word. "Research" suggests laboratories and white coats. The statute says something considerably broader, and it comes down to four questions.
1. Was there a permitted purpose?
Were you trying to create or improve a business component — a product, process, formula, invention, technique, or piece of software — in its function, performance, reliability or quality?
Note "improve." A great deal of qualifying work is not inventing something new. It is making an existing thing faster, cheaper, more durable or more consistent.
2. Was it technological in nature?
Did the work rely on the hard sciences — engineering, physics, chemistry, biology, computer science?
This does not require a PhD, a lab, or a research department. A machinist developing a fixture is applying mechanical engineering. A developer restructuring a database is applying computer science. The test is about the nature of the knowledge being used, not the job title of the person using it.
3. Were you eliminating uncertainty?
At the outset, did you know whether you could achieve the result? Did you know how you would achieve it? Did you know what the appropriate design was?
If the answer to all three was yes, it was execution, not research. If any answer was no, you had technical uncertainty — and resolving it is what the credit exists to reward.
This is the test people underestimate. Uncertainty is ordinary. It is the daily condition of anyone building something that has not been built in exactly that configuration before.
4. Was there a process of experimentation?
Did you evaluate alternatives? Modeling, simulation, systematic trial and error, testing, prototyping — anything that constitutes considering more than one approach and determining which worked.
The failed attempts count. This surprises people more than anything else on this list. The credit does not require success. A project that consumed six months and was abandoned still generated qualified expenditure, because the experimentation happened.
Where businesses actually fail
Rarely on the substance. Nearly always on the fourth test — not because the experimentation did not occur, but because nobody documented it as it happened.
Your engineers know which three approaches they tried before the fourth one worked. That knowledge lives in their heads, in Slack threads, in a whiteboard photo somebody took. An examiner three years later cannot see any of that.
That is most of what a properly run study does: capture the contemporaneous evidence of work your team was already doing, in a form that survives examination.
The practical version
If your business builds, designs, tests, formulates, codes or improves things — and sometimes those efforts do not work the first time — the question is not whether you qualify. It is how much of the work is documented well enough to claim.
That is a conversation, not a form. It costs nothing to have.