In short
- Work qualifies when the team faced a technological uncertainty, worked through it systematically, and aimed for a technological advancement.
- It is about the kind of work, not the industry. Software, manufacturing, food, engineering and life sciences all claim.
- Usually only part of a project qualifies: the experiments, not the routine build around them.
- Records made at the time, like test results, tickets and code history, are what prove it.
The idea behind the rules
SR&ED rewards one specific thing: work that tries to answer a scientific or technological question nobody could answer for you. The Income Tax Act calls it systematic investigation or search, carried out in a field of science or technology, by means of experiment or analysis. Most business claims fall under the third kind of work the law names, experimental development: creating new, or improving existing, materials, devices, products or processes.
That leaves a lot of room. You do not need a lab, a patent or a breakthrough. You need a real unknown, a planned way of working through it, and a record of what you tried. If you are new to the program, start with What is SR&ED?.
The three tests
The CRA's eligibility guidelines look at every project through three questions. A project needs a yes to all three.
- Technological uncertainty. At the start, could you know whether the goal was possible, or how to reach it, from what was publicly known or what your team already knew? The uncertainty has to be technological. Not knowing whether customers will buy it, or how long it will take, does not count.
- Systematic investigation. You worked through the uncertainty in a planned way: you formed an idea of what might work, tested it by experiment or analysis, measured the result and used it to decide the next step.
- Technological advancement. The aim was to learn something new about the technology itself. Finding out that an approach does not work is an advancement too.
The "could a professional look it up?" check
A handy way to test uncertainty: would a competent professional in the field know the answer, or find it in published sources, without experimenting? If yes, the work is probably routine engineering, however hard it was.
Examples across industries
These examples show the pattern. Each one could go either way depending on the details, which is why the records matter.
- Software
- Likely SR&EDA matching engine that must handle ten times the load within the same latency, when known designs could not
- Usually not SR&ED on its ownBuilding screens, integrating a documented API, fixing bugs with known causes
- Manufacturing
- Likely SR&EDA new process to weld a material combination that keeps cracking with standard methods
- Usually not SR&ED on its ownSetting up a production line with off-the-shelf equipment and known settings
- Food and beverage
- Likely SR&EDA plant-based formula that holds texture through freezing when known stabilizers fail
- Usually not SR&ED on its ownChanging flavours or packaging, routine shelf-life testing
- Engineering
- Likely SR&EDA mounting system for a load and temperature range no published design covers
- Usually not SR&ED on its ownApplying building codes and standard calculations
- Life sciences
- Likely SR&EDA method to purify a compound at a yield nobody has reported
- Usually not SR&ED on its ownRunning established assays for quality control
- Agriculture
- Likely SR&EDA sensor setup that predicts disease in a crop where existing models do not work
- Usually not SR&ED on its ownCollecting field data with standard methods
| Industry | Likely SR&ED | Usually not SR&ED on its own |
|---|---|---|
| Software | A matching engine that must handle ten times the load within the same latency, when known designs could not | Building screens, integrating a documented API, fixing bugs with known causes |
| Manufacturing | A new process to weld a material combination that keeps cracking with standard methods | Setting up a production line with off-the-shelf equipment and known settings |
| Food and beverage | A plant-based formula that holds texture through freezing when known stabilizers fail | Changing flavours or packaging, routine shelf-life testing |
| Engineering | A mounting system for a load and temperature range no published design covers | Applying building codes and standard calculations |
| Life sciences | A method to purify a compound at a yield nobody has reported | Running established assays for quality control |
| Agriculture | A sensor setup that predicts disease in a crop where existing models do not work | Collecting field data with standard methods |
Support work that counts
Work around the experiments can count when it is needed for, and directly supports, the SR&ED. The law lists engineering, design, operations research, mathematical analysis, computer programming, data collection, testing and psychological research.
So a developer writing the test harness for an experiment, a technician building a prototype rig, or an analyst collecting the data the experiment needs can all be part of the claim, for the time they spent on it. The same tasks done for normal production do not count.
What never counts
The definition rules out some work entirely, even if it is part of an SR&ED project:
- Market research or sales promotion.
- Quality control or routine testing.
- Research in the social sciences or the humanities.
- Prospecting, exploring or drilling for minerals, oil or gas, or producing them.
- Commercial production of a new product, or commercial use of a new process.
- Style changes.
- Routine data collection.
Drawing the line inside a project
Most real projects mix SR&ED with routine work. A new product might involve one hard technical problem and months of standard building around it. Only the part that meets the three tests, plus the support work that served it, belongs in the claim.
- Start where the team realized known methods would not do the job.
- End when the uncertainty was resolved, whether it worked or not. Work after that point, like polishing and shipping, is usually routine.
- Count time, not people. A person who spent 30% of the year on the experiments counts for 30%.
Claiming the whole project is the most common mistake
Claims that include all the work on a project tend to get cut in a review, and the cuts can take good parts of the claim with them. A smaller, well supported claim usually ends up worth more.
The records that prove it
The CRA expects evidence that the work happened the way you describe it. The best evidence is made at the time, as part of normal work:
- Notes on what you were trying to do and why known methods were not enough.
- Each idea you tested, what happened, and what you changed next, including failures.
- Test results, measurements, benchmarks and prototypes.
- Code history, design documents, tickets and meeting notes.
- Who worked on it, and their time.
You do not need special SR&ED paperwork. Most teams already create these records. The work is in connecting them to the right project and the right people, which is what Sredify does from the tools your team already uses.
Frequently asked questions
Does my work need to succeed to count as SR&ED?
No. An experiment that shows an approach does not work is still a technological advancement, as long as you were systematically testing a real uncertainty. Failed attempts are often strong evidence.
Is software development eligible for SR&ED?
Some of it. Software qualifies when the team hit a technological uncertainty that known methods, libraries and architectures could not resolve, and worked through it by testing ideas. Routine development, configuration and bug fixes with known causes do not.
Do I need a lab or a research team?
No. Many claims come from small teams doing development as part of normal business. What matters is the nature of the work, not the setting or the job titles.
Can business uncertainty count, like whether a product will sell?
No. The uncertainty has to be scientific or technological. Market, cost and schedule questions are not part of SR&ED.
Can I check eligibility before I start a project?
Yes. From April 1, 2026 the CRA offers an optional pre-claim approval, where it reviews a project's technical eligibility before you start the work or spend the money.
Prepare your SR&ED claim with Sredify
Connect the tools your team already uses, see which work may qualify, and build a claim with every hour traced to its source.
Sign in to get startedSources
- Income Tax Act s. 248(1): definition of SR&ED
- CRA: Guidelines on the eligibility of work for SR&ED tax incentives
- Budget 2025, chapter 1: pre-claim approval
- CRA: SR&ED tax incentive program
This guide explains the general rules. It is not tax advice: talk to your accountant or SR&ED preparer about your own claim.