For twenty-three years the question ran through five questions from a 1998 Tax Court decision. The CRA replaced them with two requirements. The statutory definition never moved, but a lot of the wording people still repeat comes from the old framework, and some of it was never accurate.
14 August 2026 · 11 minute read
For twenty-three years, whether work qualified for SR&ED ran through five questions. They came from Judge Bowman's decision in Northwest Hydraulic Consultants in 1998, and the CRA built them into its 2012 Eligibility of Work policy. Anyone who has been near a claim for more than a few years learned them, and a lot of published guidance still repeats them.
In the summer of 2021 the CRA replaced them. The current document is the Guidelines on the eligibility of work for SR&ED tax incentives, and it works from two requirements rather than five questions. Both frameworks were pointing at the same statutory definition, so the outcome on most files is unchanged. But the wording moved, and some of what people still repeat was never quite right in the first place.
Both requirements have to be met. Neither one carries a project on its own, and a narrative that argues only the technical difficulty is arguing half the case.
Both are drawn straight from the definition of SR&ED in subsection 248(1).
Why the work was done. It has to have been conducted for the advancement of scientific knowledge, or for the purpose of achieving a technological advancement. The guidelines put the emphasis on generating knowledge that advances the understanding of science or technology.
How the work was done. It has to have been a systematic investigation or search, carried out in a field of science or technology, by means of experiment or analysis.
The change from five questions to two is not a loosening. It is a change in where the analysis sits. The five questions walked a reviewer through uncertainty, hypothesis, procedure, advancement and record keeping in sequence. The two requirements ask the same things, but they ask them as a single judgment about purpose and a single judgment about method. A narrative organised as five numbered answers is answering a form of the question the CRA stopped asking in 2021.
This is where most weak narratives are actually lost, and it is worth being exact.
The guidelines describe uncertainty as a situation in which it is unknown, or uncertain, whether a given result or objective can be achieved, due to an insufficiency in the available scientific or technological knowledge. The operative words are the last five.
Available knowledge means two things together: what is held inside your own organisation, and what is reasonably available to you from public sources. So a problem your team could not solve, but which a competent professional in the field could have resolved by reading the literature, consulting a standard text, or applying an established technique, is not technological uncertainty. It is a gap in your knowledge rather than a gap in the field's.
In practice, establishing uncertainty properly means writing down three things: what the objective was, what was already known and tried from the existing body of knowledge, and the specific point at which that body of knowledge ran out. A narrative that asserts difficulty without doing that is asking a reviewer to take the uncertainty on faith, and reviewers are not required to.
A great deal of SR&ED advice contains some version of the line "you need a systematic investigation, not trial and error." It is repeated confidently and it is not what the guidelines say.
Iterative and trial-and-error approaches can qualify, provided they are carried out systematically. What the guidelines require of the method is a hypothesis as a starting point, testing of that hypothesis by experiment or analysis, logical conclusions drawn from the results, and evidence kept as the work progresses.
So the line is not between careful work and messy work, and it is not between a laboratory and a shop floor. It is between work in which each attempt followed from what the previous attempt showed, and work in which attempts were made without reference to the last result.
This distinction matters commercially as well as technically. It means a small manufacturer with disciplined batch records is often in a stronger position than a well-funded team that documented nothing, and it means the advice a claimant most needs is usually about recording rather than about doing anything differently.
The guidelines are unambiguous on this: success or failure in meeting your objectives is not relevant to whether the work qualifies.
What has to be attempted is an advancement in the underlying science or technology. Establishing that a particular approach cannot work, and understanding why, is an advancement in knowledge. A project abandoned halfway through the year can carry a strong claim. A product that shipped on time and sold well can carry none, if nothing about it required resolving an uncertainty.
The failure mode in Line 246 follows from this. A narrative that describes what the product now does is describing a commercial outcome. A narrative that describes what is now known, or what capability now exists and can be applied again, is describing an advancement. The two read very differently to someone assessing the claim.
The definition ends with a list of work that is outside SR&ED however difficult it was: market research and sales promotion, quality control and routine testing, research in the social sciences or humanities, prospecting and exploring for minerals, petroleum or natural gas, commercial production of a new or improved product and commercial use of a new or improved process, style changes, and routine data collection.
Support work is treated separately and conditionally. Eight categories, including engineering, design, computer programming, data collection and testing, can form part of a claim, but only where they are commensurate with the needs of, and directly in support of, the eligible work. The same activity therefore moves in or out depending on why it was performed. Testing that checks a hypothesis is support work. Identical testing performed to confirm a product meets specification is quality control, and quality control is excluded.
The two requirements decide which projects are in. A different set of rules decides what those projects are worth, and a claim can be argued perfectly on the science and still be reduced on the arithmetic.
Salaries and wages for time directly engaged in the work in Canada. Overhead through the proxy method at 55% of that salary base. Arm's length contract payments at 80% of the eligible amount. Materials consumed or transformed in the work. Capital expenditures, claimable again for spending made after 15 December 2024 following a decade out of the program.
On the credit itself, a Canadian-controlled private corporation earns 35% on qualified expenditure up to an annual limit now set at $6 million, refundable in cash, with the phase-out running from $15 million to $75 million of taxable capital. Eligible Canadian public corporations now access the enhanced rate as well. Above the limit, and for corporations outside those categories, the rate is 15%, and provincial credits sit on top with their own rates and their own rules on refundability.
The 2021 guidelines did not widen or narrow the program. They restated a statutory definition that has not moved, in language intended to be common to claimants and to CRA officials.
What that leaves is a narrower question than most advice suggests. Not whether the work was hard, not whether it was novel to the company, and not whether it succeeded. Whether there was an insufficiency in the available knowledge, and whether the work to resolve it proceeded by hypothesis, test, and conclusion, with evidence generated along the way.
Most eligible work in Canada meets both. A good deal of it is never claimed, and a good deal of what is claimed is described against a test that was retired five years ago.
If your last claim was written around the five questions, it is worth reading it against the framework that has been in force since 2021. The projects rarely change. The way the case is put usually does.
Book a meeting Check your eligibilityGeneral information current at the date of writing, not advice on a specific claim. Program rules, forms, and administrative practice change; we confirm the current position for every file.
© Sivanta Solutions Inc. · poorva@sivanta.ca · sivanta.ca