Key takeaway: Most SR&ED consultant fees are priced as a percentage of the credit you receive, not a price for the hours the work takes. That means your bill moves with the size of your claim, not with how much effort went into preparing it. For a CCPC now eligible for up to $2.1 million a year in enhanced credits under the 2026 rules, that structural difference is worth understanding before you sign anything, whatever pricing model you end up choosing.
If you've never filed a SR&ED claim before, the pricing conversation can feel like an afterthought. It shouldn't be. How a provider charges for SR&ED work shapes what happens to your claim as it grows, and growth is exactly what the 2026 rule changes were built to encourage.
This post isn't about telling you which pricing model is right. It's about giving you the math so you can work that out yourself, plus the questions worth asking before you commit a percentage of next year's refund to anyone. The same questions apply whether you're comparing traditional consultants, software-driven platforms, or handling it yourself, and the full rundown of what changed under the 2026 rules is worth reading before you price out any of them.
How SR&ED consultant fees typically work
Three broad pricing models show up across the market, and each one changes the incentive between you and whoever prepares your claim.
- Contingency pricing. You pay a percentage of the credit the CRA actually approves. Nothing upfront, no invoice if the claim is denied. The appeal is obvious: the provider's payday depends on your claim succeeding, so it looks aligned. The catch is what that percentage is a cut of. It's a cut of the outcome, not a price for the work, and those two things don't move together.
- Flat-fee pricing. You pay a fixed price set by the scope of the engagement, regardless of how large the resulting credit turns out to be. The incentive shifts: the provider is paid to do the work well, not to maximize a percentage of your refund.
- Hourly or project-based billing. Less common for SR&ED specifically, but some accounting firms fold SR&ED prep into general advisory billing. You pay for time spent, win or lose, which puts the underclaiming risk on you if nobody has a strong incentive to dig for eligible work.
None of these models is inherently dishonest, and each can produce a well-documented, defensible claim. What's worth doing the arithmetic on is what each model costs you as your claim gets bigger, because under the 2026 changes, claims are getting bigger by design.
The contingency math as your claim grows
Here's the structural point, stripped of any specific rate, using a hypothetical to show the shape of it rather than a quoted price.
Say a provider charges on contingency. Whatever the exact number, it's applied to the credit amount, not the hours worked. If your claim earns a $400,000 credit, the fee is some fraction of $400,000. If next year your R&D spend grows and the same claim earns $1.2 million, the fee is the same fraction of $1.2 million; three times the dollar cost, for work that may not have taken three times the hours. The percentage didn't change. The bill did, because the bill was never tied to effort in the first place.
That gap is bigger than it used to be, because the ceiling moved. Under the Budget 2025 changes that became law when Bill C-15 received royal assent on March 26, 2026, the enhanced 35% refundable credit now applies to up to $6 million of qualifying expenditures per year for CCPCs, double the old $3 million limit. At the full limit, that's up to $2.1 million in credits in a single year, refundable on current expenditures with zero tax owing. A company that grows into that limit isn't just claiming a bigger credit. On a contingency structure, it's paying a proportionally bigger fee for work that, past a certain point, doesn't get proportionally harder. Assembling a T661 for $6 million of qualifying spend is more work than for $1 million, but it isn't six times the work.
Flat-fee pricing doesn't have this problem by construction. The price is set by the scope, the number of projects, the complexity of the technical narrative, the volume of documentation to review. If your R&D spend doubles next year without the underlying work getting materially more complex to document, a flat fee doesn't automatically double with it.
This isn't an argument that contingency pricing is a bad deal in every case. For a company with no cash to pay upfront and real uncertainty about whether the technical work will hold up, "no claim, no fee" removes a real risk. The point is narrower: know which variable your fee is actually a function of before you agree to it.
This is where the math bites hardest for software companies specifically, since software development took 42.6% of all SR&ED credits allowed in FY2025-26, per the CRA's annual program statistics — the largest share of any sector, and the sector most likely to be growing its R&D spend year over year as headcount and product surface area grow. A software team that doubled its eligible spend because it doubled its engineering headcount didn't necessarily double the number of distinct technical narratives that need writing. Under contingency pricing, the bill doubles anyway.
Consultants, software, or both — what the model actually changes
Pricing model and delivery model are two separate decisions, and it's worth untangling them before you shop. A traditional consulting engagement is usually a person or small team who interviews your engineers, writes the technical narrative, and assembles the forms, priced either on contingency or flat fee. A software-driven platform usually captures evidence continuously from your existing tools — commits, tickets, time tracking — and generates the narrative from that record rather than from a year-end interview, which changes when the documentation is created, not just who's paid to create it.
Neither delivery model automatically fixes the incentive problem described above. A software platform priced on contingency has the same scaling issue as a human consultant priced the same way: the bill still moves with the credit, not the work. Conversely, a traditional consultant on a flat fee who nonetheless builds the narrative from real-time client interviews throughout the year, rather than a single year-end session, can produce documentation just as strong as any platform's. The variable that actually predicts claim quality is when the record was created relative to when the work happened, not whether the word "software" or "consultant" appears on the invoice.
What a flat fee actually protects you from
The version of this argument that matters isn't abstract. It's about what happens at renewal, two or three years into a relationship, once your R&D spend has grown past whatever it was when you first signed.
Under a contingency arrangement, the provider has no built-in reason to ever revisit the percentage as your claim scales, because their revenue is already scaling with it automatically. Under a flat-fee arrangement, the incentive runs the other way: if the scope of the work genuinely grows, the provider has to justify a new price, which means someone is forced to actually look at whether the work grew or just the dollar figure did.
There's a second, quieter effect worth naming. A percentage-of-refund structure creates a soft incentive to find every possible dollar to include, whether or not it's the strongest-documented dollar. That's not necessarily a technical-eligibility problem, since the CRA's eligibility test is the same regardless of who prepares the claim. It's a documentation-quality problem: a claim padded to maximize the percentage cut is a different object than a claim built to survive scrutiny. And per the CRA's own annual program statistics, the FY2025-26 numbers show 90% of processed claims accepted exactly as filed, 6% accepted after modification, and 4% denied. In our experience — and this is our read, not the CRA's — the claims that land in that last group are rarely wrong about the science. They're weak on the record of it. Fee structure doesn't determine which side of that line you land on, but it does determine who's incentivized to care.
Why bigger claims don't need bigger risk
It's tempting to assume a larger claim under the new $6 million limit automatically means more audit exposure, and therefore more reason to lean on a provider who "handles everything." The CRA's own numbers don't support the first half of that assumption. Review outcomes aren't published broken out by claim size, but the standards that govern processing are size-agnostic: refundable claims accepted as filed are processed within 60 calendar days, and the CRA hit that standard 95% of the time last fiscal year, regardless of how large the credit was.
What actually drives whether a claim survives a CRA review is the quality of the documentation behind it, not the dollar amount on the T661 or who signed the engagement letter. A $2 million claim with contemporaneous technical records, timesheets tied to specific uncertainties, and a narrative written by people who did the work is a stronger claim than a $200,000 one reconstructed from memory eighteen months later. Fee structure is a proxy for incentive alignment, not for claim quality. The actual driver of claim quality is when the documentation was created relative to when the work happened.
That's the honest counterpoint to the entire "which pricing model is better" framing: pricing model is a second-order question. The first-order question is whether whoever prepares your claim, at whatever price, is building it from real-time records or backfilling it at deadline. Get that part right and the fee structure argument mostly becomes about cash flow timing, not risk.
Doing it yourself vs. paying anyone at all
Not every company needs to pay a consultant, a software platform, or anyone else for SR&ED prep, and it's worth saying so plainly. If your claim is small, your projects are simple to describe, and someone on your team already understands the eligibility test well enough to write a defensible technical narrative, filing it yourself is a legitimate option. The CRA doesn't require a professional preparer. You're free to fill out Form T661 and the relevant schedule on your own, and plenty of companies do.
The case for paying someone, in whatever form, gets stronger as three things grow at the same time: the dollar value at stake, the number of distinct projects that need separate technical narratives, and the distance between the people who did the engineering and the people writing the claim. A five-person startup with one clear R&D thread and a founder who can write the narrative doesn't have the same calculus as a fifty-person engineering org running six workstreams where the person who solved the hardest problem left the company eight months ago.
One more honest point: whoever you pay, you're still the one legally responsible for the claim. The CRA is explicit that the claimant is accountable for the claim's completeness and accuracy whether they prepared it themselves or paid someone else to. "We paid a firm to handle it" isn't a defense if the documentation doesn't hold up. That's a reason to care about the fee structure's incentives in the first place, not just its price.
Questions worth asking before you sign
Regardless of which model a provider quotes you, a few questions cut through the pricing pitch:
- What happens to the fee if my claim doubles next year without the underlying work getting materially harder to document? A straight answer here tells you a lot about which variable the price is actually tied to.
- Who writes the technical narrative, and when? If it's assembled at filing time from a founder's memory rather than from records kept while the work happened, that's a documentation risk no fee structure fixes.
- What happens if the CRA selects the claim for review? Ask specifically whether support continues at the quoted price or becomes a new, separate engagement once a review starts.
- Is the person accountable for the technical eligibility call independent of the person selling the engagement? Their incentives should point at claim quality, not just at closing the deal.
- Does the fee change if part of the claim is denied or modified? On a pure contingency model, a partial denial should lower the fee proportionally; confirm that's actually how it works before signing.
None of these questions require you to know SR&ED law. They're about incentives, and incentives are exactly what a fee structure sets.
Glauq's own model is flat-fee for this reason: the price is set by the scope of the engagement, not by the size of the credit, so a bigger claim under the new $6 million limit doesn't mean a bigger bill for us to earn. Evidence is captured continuously from the tools your team already uses, and a qualified, independent SR&ED expert reviews and stands behind every claim before it's filed. That's one way to structure the incentive; it isn't the only legitimate one, and it won't be the right fit for every company.
Frequently asked questions
Is a contingency SR&ED fee ever the better choice? It can be, particularly for a company with real cash constraints or genuine uncertainty about whether its work will qualify, since "no claim, no fee" removes the upfront-cost risk. The trade-off is that the fee scales with the credit amount rather than the work involved, which matters more as your claim grows under the $6 million expenditure limit.
Does the CRA regulate what SR&ED preparers can charge? No. Fee arrangements between a claimant and whoever prepares their claim, whether a consultant, an accountant, or a software provider, are a private commercial matter. The CRA's rules govern eligibility and filing requirements, not pricing.
Am I still responsible for my claim if I pay someone else to prepare it? Yes. Per the CRA's submission guidance, you're responsible for your claim being complete, accurate, and supported by evidence whether you prepared it yourself or paid someone else to do it.
Can I file SR&ED without paying anyone? Yes. There's no requirement to use a professional preparer. Filing Form T661 and the applicable schedule yourself is a legitimate path, particularly for a small, single-project claim with straightforward documentation already in hand.
Does a bigger claim under the 2026 rules mean a higher audit risk? The CRA doesn't publish review rates by claim size, and its processing service standards apply regardless of claim size: 60 calendar days for accepted-as-filed refundable claims, hit 95% of the time in FY2025-26. What drives review outcomes, in our experience, is documentation quality, not the dollar amount on the form.
What's the difference between a contingency fee and the prescribed proxy amount? They're unrelated. The prescribed proxy amount is a CRA-set rate, 55% of your salary base, used to estimate overhead costs inside your claim's expenditures. A contingency fee is what you separately pay whoever prepares the claim; it has nothing to do with the proxy calculation.
Whatever pricing model you choose, the number that actually protects your claim isn't the fee. It's whether the documentation behind it was built while the work happened or reconstructed after the fact.
See what your claim could be worth — estimate your refund or check your eligibility.