Recover up to 64% of your R&D spend

SR&ED is a tax credit, not an envelope you apply for. Nobody awards it: you claim it with your tax return, after the fact, for work already carried out. There is no competition, no selection committee and no envelope running dry — but no discretionary goodwill either: a documented claim goes through, an undocumented one is denied. A grant works the other way around: you apply BEFORE incurring the expense and you are judged on the application. The two stack, with one caveat that matters — a grant received is government assistance, and it is deducted from the SR&ED expenditure base.

Work that sets out to resolve a technological uncertainty through a systematic investigation. Three conditions: a technical obstacle whose solution was neither known nor deducible from standard practice; hypotheses formulated, tested, and whose results are recorded; an advancement in knowledge, even a negative one — a failed experiment still qualifies if it taught you something. What does not count: commercial novelty, the size of the budget, the success of the product. A project that is unremarkable to the market but technically uncertain for your team qualifies; a spectacular product assembled from known techniques does not.

Salaries of staff directly engaged in the work come first: it is the heaviest line, often most of the base. Then subcontracting, only partly retained, materials consumed or transformed during testing, and overhead — usually through the proxy method, 55% of eligible salaries, rather than itemised line by line. Since the CRIC, Québec claims must also account for acquired property, land and buildings excepted, and for pre-commercialisation work. What is almost always left out: owner time spent on technical problems, and support staff assigned to the trials.

Evidence that is contemporaneous with the work. The CRA is not asking for a report written after the fact; it is asking for the trace of what happened during: technical meeting notes, tickets and version history, lab notebooks, test protocols and results, successive drawings, time records by project. A timesheet kept every week carries more weight than a thirty-page memo written eighteen months later. The habit that changes everything: record the hypothesis and the result at the moment the test happens — including, and especially, when it fails.

We identify the eligible projects with your technical teams, structure the evidence, compute the expenditure base for both levels in the right order, write the technical description of the work and assemble the claim with your accountant or tax adviser. If the CRA or Revenu Québec opens a review, we take the audit on and defend the file.

Yes, and it is the rule rather than the exception: NRC-IRAP, ESSOR, the C3i or a sector-specific programme all stack with SR&ED. Two caveats. First, government assistance received is deducted from the SR&ED base: it does not cost you the credit, it shrinks its base — stacking always pays, just less than the sum of the headline rates. Second, in Québec the CRIC is exclusive: one expenditure cannot carry both the CRIC and another Québec credit, which forces a choice, typically between the CRIC and the C3i on an equipment purchase. The order in which claims are built changes the total collected.

Yes, but not on the grounds that it is software. What qualifies is the portion of the work where an engineering problem had no known solution: an algorithm whose performance broke down at the target scale, a distributed architecture under latency or consistency constraints, a machine-learning model whose results were not predictable, an integration that forced you around an undocumented limitation. What does not qualify: integrating a library, wiring up an API, redesigning an interface, configuring off-the-shelf software. It is the sector the CRA audits most, which is why R&D work should be separated from routine development right in the time tracking.