Key Takeaway
SR&ED claims are won or lost on documentation quality and hitting the 18-month deadline. Get these five steps right and you claim more while staying defensible.
SR&ED (Scientific Research and Experimental Development) is Canada’s largest R&D funding program, and the filing trips up more companies than the eligibility rules do.
Five steps decide whether a claim holds. This article covers each one, what the CRA actually wants at each stage, and the deadline that ends the conversation if you miss it.
Step 1: Is your work eligible?
Check this before you spend a week on documentation. Eligible work involves systematic investigation that resolves a technological uncertainty and advances technical knowledge.
What qualifies: developing new algorithms or software architectures, creating new materials or manufacturing processes, resolving technological uncertainty through experimentation, engineering work that advances technical knowledge, and experimental development of new products or processes.
What doesn’t: routine testing and quality control, market research, aesthetic and style changes, data collection unrelated to research, and routine software maintenance.
Four questions per project
Run each project through these. If you can’t answer yes to the first three, it isn’t SR&ED.
- Was there a technological uncertainty you couldn’t resolve with standard practice?
- Did you investigate it systematically rather than by trial and error?
- Did the work advance technical knowledge?
- Does that advancement apply beyond this one project?
The first question is where most claims fail. Difficulty isn’t uncertainty. A hard problem with a known solution doesn’t qualify.
Step 2: What do you need to document?
Documentation is the foundation, and the CRA weighs it more heavily than the technical merit of the work.
Project records. Objectives and hypotheses. Experimental designs and test plans. Results, including failures. Iterations and what changed between them. Conclusions.
Personnel records. Who worked on each project, their qualifications and role, time spent on eligible activity, and the nature of their contribution.
Financial records. Salaries and wages for R&D staff, materials consumed, contractor costs, overhead and equipment.
Five practices that make documentation hold
Write it as you go. Records created during the work carry weight. Records assembled at filing time invite questions.
Date everything. Timelines are what let a reviewer verify the work happened when you say it did.
Keep the failures. Failed experiments are usually your strongest evidence, because a dead end is hard to invent and proves the outcome was uncertain.
Be specific. “Improved performance” supports nothing. Name the constraint and the measurement.
Preserve the artifacts. Lab notebooks, commits, design documents, meeting notes. These are what a narrative points back to.
Step 3: Which expenses can you claim?
Personnel. Salaries and wages for people directly engaged in SR&ED, plus benefits, employer contributions and bonuses attributable to the R&D work. This is the biggest line for most software companies.
Materials. Raw materials consumed or transformed during the work, prototype materials, testing supplies.
Contractors. Payments to arm’s-length contractors performing SR&ED, technical consultants, and testing services.
Equipment. Leases for R&D purposes, and depreciation in some cases.
Allocating shared resources
Most people don’t work on SR&ED full time, which makes allocation the part reviewers examine hardest.
Keep detailed time records. Use one allocation methodology and apply it consistently. Write down the basis for each allocation. Then be ready to defend it, because “roughly 70%” is not a methodology.
Step 4: How do you write the technical narrative?
Assume your reviewer is technically literate but not an expert in your field. The narrative has to make the uncertainty legible to them.
Objectives. What you were trying to achieve, and why existing solutions were inadequate. The second half matters more than the first.
Technological uncertainties. The specific challenges that standard engineering and available knowledge couldn’t resolve. Be concrete: name the constraint, the threshold, the failure mode.
Methodology. Your systematic approach. Hypotheses tested, experiments run, variables controlled, analysis methods used.
Advancements. What new knowledge or capability came out of it, including when the project failed. Learning that an approach doesn’t work is an advancement.
Writing tips
Use plain technical language, not marketing language. Focus on the technical content rather than the business benefit, because the business benefit is irrelevant to eligibility. Expand acronyms. Tie every activity to a specific uncertainty. And show the iteration, because that’s what distinguishes investigation from implementation.
Step 5: When do you have to file?
Within 18 months of your fiscal year-end. The CRA enforces this strictly. There is no late filing.
What the filing requires
The T661 is the primary form. It needs project descriptions, expenditure details, supporting calculations and certification by an authorized person.
Your claim amounts also have to reconcile against payroll records, accounts payable, general ledger entries and your tax filings. Reviewers check this, and a mismatch is an easy reason to look harder at everything else.
Should you hire a specialist?
Usually yes, at least for your first claim. A specialist reviews projects for eligibility, prepares the technical narratives and handles CRA questions.
Fees commonly run 10-30% of claim value on contingency, depending on the firm and the size of the claim, which means you pay only if the claim succeeds. For larger claims, a flat fee is normally cheaper. We compare the options in 17 R&D tax credit specialists in Toronto.
What this actually buys you
Credits, obviously. But the companies that do this well report a second benefit that shows up by year two.
The discipline of recording hypotheses, failures and time allocation is the same discipline that makes project retrospectives useful and estimates accurate. You end up running R&D better, not just claiming it better.
Frequently Asked Questions
What happens if I miss the 18-month deadline?
You lose those credits for that fiscal year. The CRA does not accept late claims. Set a reminder at least three months before the deadline so preparation isn’t rushed.
Can failed projects still qualify?
Yes, and they often make stronger claims. The CRA cares about technological uncertainty and systematic investigation, not commercial success. A failed experiment demonstrates the uncertainty existed, which is exactly what you have to prove.
How far back can I claim if I’ve never filed?
Up to 18 months from the end of each fiscal year. Anything older is gone. If you’ve never claimed, start documenting now so the current year is complete, and check whether the most recent year is still within the window.
Need help with your next filing? Talk to our team about building the documentation before the deadline rather than after it.