Canada gives back a share of what you spend building software. Most founders miss it because it arrives late.
Software development took 42.6 percent of the research and development tax credits Canada allowed last year, three times the next largest field and a bigger share every year for four years running. Kevin Barton spends his week talking to the founders doing that work. Most of them have never heard of the program.
His explanation is not that it is obscure. It is that it pays at the wrong moment to be noticed.
The credit is claimed on work already finished, and it arrives months after a year end, long after the founder has stopped thinking about the build it covers. What follows is what the money is, why so much of it goes unclaimed, and what a founder can settle before another year closes.
Two Things Founders Assume, Both Wrong
The Scientific Research and Experimental Development program is the largest federal support for research and development in Canada. In the year to 31 March 2026 the Canada Revenue Agency allowed $4.6 billion in investment tax credits across 24,160 claims filed. It approved 90 percent of them as filed.
The first assumption is that it is a specialist scheme for laboratories. Software development took the largest single share of those credits, roughly three times the next field on the list.
The second is that it is for bigger companies. Businesses making under $4 million a year filed 65 percent of processed claims and received a third of the credits allowed.
Kevin takes three to five founder calls a day, and the same gap comes up on most of them.
Most founders either haven’t heard of it or assume it’s for somebody bigger. It’s retroactive, so it lands well after the work is done, which is exactly why people write it off. Then a year of development goes by and nothing gets claimed.
What a Claim Actually Has to Say
A claim is not a receipt for money spent on engineers. It has to identify a technological uncertainty the team could not resolve from what was already known, and describe the systematic work done to resolve it, separated from the routine development around it. Kevin describes the core of the job as translation: finding the real technical risk inside a build and putting it in the terms the program asks for.
Two practical details decide most claims before the writing starts. The documentation has to be contemporaneous, made while the work was happening rather than reconstructed a year later. And the reporting deadline is 18 months after the corporation’s year end. An expenditure not reported by then cannot earn the incentive at all.
The program has also just widened. Bill C-15 received Royal Assent on 26 March 2026 and doubled the annual expenditure limit for the enhanced 35 percent refundable credit from $3 million to $6 million, which lifts the maximum annual refundable credit from $1.05 million to $2.1 million. The taxable capital phase-out band moved from $10 million and $50 million to $15 million and $75 million, so companies that had grown out of the enhanced rate are back inside it, and capital expenditures are eligible again. The changes apply to tax years beginning on or after 16 December 2024, which means the year most founders are about to file for.
What Happens When It Gets Handed Off
Thrive started doing the claim work inside its engagements after watching what happened when it did not.
We know that money has to be spent with us anyway, so there’s no point kicking you off to a consultancy or leaving it on your own. What we found is a lot of our clients then just don’t do it. It’s confusing and they have other priorities. So that’s why we just include it. It’s not billable.
The competitive version of the same handoff has a cost attached. “Every agency we compete with will happily bill you for a build and then let you figure out the funding side yourself,” Kevin has written, “or hand you off to a consultant who takes a cut of whatever comes back.”
There is a practical reason to keep the claim near the engineering rather than near the accounting. The people best placed to describe a technological uncertainty are the ones who worked through it. A claim assembled a year later by somebody who was not in the room has to reconstruct what the team already knew, and has to do it from documentation nobody wrote for that purpose.
Thrive identifies what applies, documents the work and guides the claim, at no separate fee, with the proceeds going to the client. To date the firm has helped clients access more than $1.2 million in funding.
The Man Who Sat at the End of the Queue
The instinct to look at where a process fails rather than where it delivers came from a job most people leave as fast as they can.
The QA years were in the early days of a Calgary cybersecurity startup, before it became Passportal, before SolarWinds acquired it and before N-able was spun out. Early stage means everyone wears several hats, and QA in particular sits at a strange vantage point. You see every business process in the company, most of which people never get to look at, let alone work inside.
What he kept seeing from that seat were features arriving at the end of the queue that missed what partners needed, months after the misunderstanding that caused them. So he moved to product management to sit at the start of the process instead of the end. The failure that recurred most was onboarding: the product worked, the demo was excellent, and then a real customer had to configure it and the whole thing stalled. Not a hard problem, as he puts it, just nobody’s job.
The funding claim is the same shape of problem. It is not difficult and it is not contested. It simply belongs to nobody, so it does not happen. “From QA I learned that the expensive mistakes are the ones nobody was incentivised to mention early,” Kevin says, “so we took the incentive out.”
It is a habit he still goes looking for in other people. When Ian Percival, Director of Software Engineering at Highwood Emissions, spoke at a Calgary meetup on handing the definition of “correct” back to the domain experts who own it, Kevin was in the room, calling it a talk worth showing up for.
One founder who came to Thrive had already looked at three other development shops. What decided it was not the rate. It was that the funding work sat inside the engagement rather than being something he would have to go and organise himself.
Where Kevin Is Now
He is in Calgary, working with teams across Canada and the United States, and spending most of his week on intake, understanding the problem behind a request before anyone writes code against it.
His standing invitation to founders is a short call with no deck attached, and he gives real advice on it whether or not they ever sign.
His test for whether a build is set up correctly has nothing to do with the plan. “In fifteen years of building software, I’ve never seen a project where what we knew at kickoff was still intact by month four,” he wrote recently. “The real question is whether you’re set up to act on it the same week you learn it.”
The Kevin Barton Playbook: Before Another Year Closes Unclaimed
- Ask well before year end. The claim covers work already done, and the details are far easier to capture while the work is still fresh.
- Separate the research from the routine. A strong claim names the uncertainty inside a build rather than describing everything that got built.
- Write it down as it happens. Documentation has to be contemporaneous, so a reconstruction assembled at filing time is the weakest version of your own case.
- Check what the new limit does to you. The enhanced rate now runs to $6 million and the phase-out band has widened, so companies that were outside it may be inside it again.
- Keep the engineers near the claim. The people who worked through the technical risk are the only ones who can describe it accurately.
Kevin Barton is Co-Founder and Chief Operating Officer of Thrive Creative Inc.
From the Diary of Kevin Barton
Kevin Barton is a Calgary-based product and operations leader and the Co-Founder and Chief Operating Officer of Thrive Creative Inc. Thrive Creative provides Development Department as a Service, an on-demand engineering department of developers, designers, data and AI specialists and project leads that a business plugs in as the work requires, with no minimums and full ownership of the code. He worked in software quality assurance, test engineering and product management across seven years through Passportal and SolarWinds and four at N-able, then a product leadership role at CyberQP, and he holds the Agile Certified Product Manager and Product Owner credential from Pragmatic Institute. On putting the funding work inside every engagement he says: “It’s the part nobody believes until they see it in the agreement.”
