Can you claim SR&ED if your project failed?

Yes. You can claim SR&ED for a project that failed, and CRA says so in plain language: "success or failure in meeting your objectives is not relevant when assessing whether your work meets the 'Why' requirement" (Guidelines on the eligibility of work). If your team spent last year chasing something that did not work, that is not a disqualifier. It may be the most claimable work you did.

This is the thing Canadian founders get wrong most often about SR&ED. The intuition is reasonable. You are asking for public money, so surely you should have something to show for it. That is not the test. The test is why you did the work and how you did it, and a project you eventually killed can satisfy both.

At a glance

• "Success or failure in meeting your objectives is not relevant" to the "Why" requirement (CRA guidelines, dated August 13, 2021).
• Rejecting a hypothesis can itself be an advancement, "because it eliminates a possible solution" (T4088, Line 246 guidance, updated February 18, 2026).
• "You do not have to achieve your goal in order to gain new knowledge" (CRA, what work is eligible, updated April 1, 2026).
• T661 Line 204 accepts a completion date because you "terminated the project for any reason" (T4088).
• What does disqualify you: working around the uncertainty instead of addressing it, and having no evidence (CRA guidelines).

Why CRA does not care whether it worked

The SR&ED definition turns on purpose and method, not results. CRA asks whether the work was done to achieve a scientific or technological advancement against a real uncertainty (the "Why"), and whether it was a systematic investigation that tested a hypothesis by experiment or analysis (the "How") (CRA, what work is eligible). Neither question asks what happened at the end.

The guidance goes further than merely tolerating failure. On Line 246 of the T661, CRA writes that "failure to achieve your objectives or goals does not necessarily mean there is no advancement," and that "the rejection of a hypothesis can be considered as an advancement because it eliminates a possible solution" (T4088). Learning that an approach is a dead end is knowledge the field did not have before you spent the money finding out. That is the whole logic of the program.

A failed project is often the stronger claim

This sounds like consolation. It is closer to a structural fact about how the test works.

If your project succeeded on the first attempt with known tools, you have a hard problem on your hands: showing that the knowledge base was ever insufficient. CRA is explicit that a novel, more capable product does not settle the question, warning that "the novelty, innovation, uniqueness, feature enhancement, or increased functionality of the product or process may not in itself demonstrate scientific or technological advancement" (T4088). A great product is not the same thing as an advancement.

The failed project usually has what the successful one lacks. It has a documented uncertainty, because you hit a wall. It has hypotheses, because you had to guess at ways around it. It has tests and logical conclusions, because you ran them and they told you no. Those dead ends are not embarrassing padding to be edited out. They are the evidence.

What actually disqualifies the work

Not failure. The disqualifier is going around the problem instead of at it. CRA puts it directly: "When faced with such an uncertainty, if you do not address it, or you circumvent it with a workaround using available knowledge, then the work will not be eligible" (CRA guidelines). The team that hit the wall and shipped a workaround has a smaller claim than the team that hit the wall and lost to it.

Buying the answer does not count either. CRA states that "acquiring available knowledge or know-how, for example, through training, on-the-job learning, hiring expert employees or consultants, or by purchasing proprietary knowledge does not meet the requirement of attempting to achieve technological advancement." If someone out there already knew, and you could reach them, there was no uncertainty to resolve.

One more trap sits in the word "systematic." Founders hear "systematic investigation" and think of their sprints, their code review, their CI pipeline. CRA closes that door before you open it: "it is not enough that work be carried out systematically in order for it to be eligible." Disciplined execution of established methods is not an investigation. A tidy team with no hypothesis can fail this test while a scrappy one with a real hypothesis passes it.

The real risk with a dead project is the evidence

Here is where failure genuinely does cost you something, though not in the way founders expect. Eligibility survives a dead project. The proof usually does not.

CRA is blunt about what happens when records are missing: "Work for which you have no relevant supporting evidence will likely be disallowed," and "contemporaneous documentation that is dated, signed, and specific to the work performed are the best supporting evidence that you can provide" (T4088, Appendix 2).

The trap is behavioural rather than technical. Teams curate their winners and bury their losers. When a project dies, the Slack channel gets archived, the branch gets deleted, the design doc goes to the bottom of a drive nobody opens, and the two engineers who ran the experiments leave in the spring. Eleven months later you sit down to write the claim and the most eligible work you did is the work you can least prove.

If you are claiming for the first time and this is making you nervous, note that CRA itself says "the lack of detailed documentary information should not discourage you from making an SR&ED claim or be considered as an indication that SR&ED did not take place, particularly in the case of a first-time claimant." Thin records are a reason to claim carefully. They are not a reason to skip it.

What goes on the form when you killed the project

The T661 has a field for this. Line 204 asks for a completion date, and CRA defines that as the point at which you "achieved the scientific or technological advancement," or "determined that the scientific or technological uncertainties cannot be resolved," or "terminated the project for any reason" (T4088). Concluding that it cannot be done is a completed project, not a failed application.

Two details follow from that. CRA notes that "the eligibility of the work also ceases at that point," so effort spent after the kill decision is not claimable even if people were still tidying up. And the claim is annual: Line 244 asks what work you performed "in the tax year," so a project you abandoned in month seven is an ordinary claim covering the months it ran.

Frequently asked questions

Can you claim SR&ED if your project failed?

Yes. CRA states that success or failure in meeting your objectives is not relevant to whether the work meets the "Why" requirement. Eligibility depends on whether you were pursuing an advancement against a real uncertainty and investigated it systematically.

Does a failed experiment count as an advancement?

It can. CRA's guidance for Line 246 says the rejection of a hypothesis can be an advancement because it eliminates a possible solution. Learning that an approach does not work is new knowledge.

We abandoned the project halfway through the year. Can we still claim it?

The T661 treats termination as a normal completion event, and Line 204 accepts a completion date because you terminated the project for any reason. Eligibility ceases at that date, and claims are filed per tax year, so you describe the work performed during the year up to that point.

Does a failed project still need documentation?

Yes, and this is the part that bites. CRA says work with no relevant supporting evidence will likely be disallowed. Dated records made while the work happened are what carry the claim, and dead projects are the ones teams tend to stop documenting.

If failure does not disqualify us, what does?

Circumventing the uncertainty with a workaround using available knowledge, or acquiring the answer through training, hiring an expert who already knows it, or buying proprietary knowledge. In both cases there was no uncertainty left to resolve.

The part worth remembering

The program is not a prize for shipping. It exists because someone has to pay for the experiments that do not pan out, and CRA has written that into the guidance more plainly than most founders realise. Your dead project is eligible work. Whether it is a claim depends on whether you can still prove it happened, which is an argument for capturing evidence while the work is live rather than reconstructing it later. That is the problem SREDlog was built around, though the habit matters more than the tool. Archive the failed branch. Keep the design doc. The experiment that told you no is worth money, but only if you can show it ran.