Most founders lodging an R&D Tax Incentive claim spend their energy on the wrong question. They ask whether their work counts as R&D. The better question is how their work splits between core R&D activities and supporting R&D activities, because that split is the single biggest driver of both how much you get back and how well your claim holds up if AusIndustry or the ATO takes a closer look.
Get it right and a legitimate claim sails through. Get it wrong and you either leave money on the table by classifying real experimental work as support, or you overreach by dressing up routine work as core and hand a reviewer an easy reason to cut your claim down.
Here is how the two categories actually work, in plain English, with the traps we see most often.
Why the split exists at all
The R&D Tax Incentive is jointly run by AusIndustry, which decides whether your activities qualify as R&D, and the ATO, which administers the offset and pays it. Every registration you lodge asks you to describe your activities under two headings.
For companies with aggregated turnover under $20 million, the program refunds up to 43.5% of eligible R&D spend as cash, even if the company is loss making. Companies at $20 million and above get a non-refundable offset of between 38.5% and 46.5%. You need at least $20,000 of eligible R&D expenditure to claim at all.
The two categories are not interchangeable. Core activities are the experiments. Supporting activities are the work that feeds them. Both attract the same offset rate on eligible spend, but supporting activities only qualify because a core activity exists to support. No core activity, no claim. That is why the classification matters so much.
What makes an activity a core R&D activity
The legislation, at section 355-25 of the Income Tax Assessment Act 1997, sets a three part test. All three parts have to be satisfied.
The outcome could not be known in advance
The result of the activity has to be genuinely unknown, and unknowable, based on current knowledge, information or experience. Not unknown to you. Unknown to a competent professional in the field.
This is where most founders sell themselves short. They think “someone somewhere has probably solved this” and assume they are out. But if the solution is not publicly available, not documented, and cannot be worked out by applying known methods, the uncertainty is real.
The practical test we use on eligibility calls: was there a point where you genuinely did not know whether it would work? Did you try things that failed? Did you have to change approach and go again? That pattern is what technical uncertainty looks like day to day.
The work followed a systematic progression
The activity has to move from hypothesis, to experiment, to observation and evaluation, and then to logical conclusions, based on principles of established science.
That sounds like a lab coat requirement. It is not. A team that writes a design doc setting out an approach, builds it, benchmarks it against a target, finds it fails at a certain load, and revises the architecture has run exactly that progression. The paperwork just happens to be a Notion page and a set of pull requests rather than a lab notebook.
What does not satisfy this test is trial and error with no framework. Randomly changing settings until something works is not a systematic progression. There has to be a reason you tried what you tried.
The purpose was to generate new knowledge
The activity has to be conducted for the purpose of generating new knowledge, including new or improved materials, products, devices, processes or services.
Note what this does not require. It does not require the knowledge to be world changing, patentable, or even ultimately useful. Failed experiments generate new knowledge. Plenty of our clients’ strongest core activities are the ones that did not work out.
It also does not require the new knowledge to be your only purpose. Commercial motivation is entirely fine. You are allowed to want to make money.
What counts as a supporting R&D activity
Section 355-30 covers supporting activities. These are activities directly related to core R&D activities. Think of the scaffolding around the experiment.
Common examples in a software or hardware business:
Building the test environment or rig your experiment runs on.
Collecting, cleaning and labelling the dataset a model is trained against.
Literature and prior art reviews done to frame the hypothesis.
Project management and technical coordination specifically for the R&D work.
Producing a prototype for the purpose of testing, as distinct from producing it to sell.
The dominant purpose rule is where claims come unstuck
Here is the catch that trips up more claims than anything else. For most supporting activities, “directly related” is enough. But if the activity produces goods or services, or falls into one of the excluded activity categories, it only qualifies if it was undertaken for the dominant purpose of supporting a core R&D activity.
Dominant purpose means the main purpose, not one of several. If you built a production feature and happened to learn something along the way, the dominant purpose was shipping the feature, not supporting an experiment. That activity is out.
This is the area where AusIndustry and the ATO push back hardest, and it is the reason a claim built by someone who understands the distinction looks very different from one built by someone guessing.
Where founders get the split wrong
Registering everything as core. A registration where the entire engineering year is described as core experimental work reads as unreliable, because no team spends every hour on experiments. It invites scrutiny you do not need.
Registering everything as supporting. The opposite mistake, usually from founders who are nervous about overclaiming. If there is no core activity, there is nothing for the support to attach to and the claim fails on its face.
Describing the project rather than the activity. “Built the platform” is a project. The registration asks for activities. A single project can contain two or three genuine core activities and a handful of supporting ones. Lumping them into one block loses the detail that makes the claim defensible.
Claiming offshore work as core. Overseas development spend does not qualify without an Overseas Finding, which has to be applied for in advance. Australian spend on founder time, local contractors and Australian based tooling can still support a solid claim, which is why it is always worth checking rather than assuming you are out.
Bundling in the excluded categories. Market research, market testing, routine quality control, cosmetic user interface changes, standard software builds and management studies are all specifically excluded as core activities. Some can still qualify as supporting activities, but only under the dominant purpose test.
The records that back the split up
The program expects contemporaneous records. That means created at the time, not reconstructed after the fact from memory.
Nobody we work with keeps perfect timesheets, and that is fine. What actually gets used:
Git commit history and pull request discussions, which are timestamped and show iteration.
Design docs, architecture decision records and technical specs.
Sprint boards, tickets and their comment threads.
Meeting notes, Slack threads and emails where technical decisions were argued out.
Test results, benchmark outputs and error logs.
Invoices and payroll records to tie the spend to the people doing the work.
The pattern a reviewer wants to see is straightforward. Here is what we did not know. Here is what we tried. Here is what happened. Here is what we did next. If your records tell that story, your split between core and supporting is easy to justify.
What changes from 1 July 2028
The 2026 Budget announced a significant reshape of the R&D Tax Incentive. The headline items are a single core rate of around 48%, the refundable threshold lifting from $20 million to $50 million of turnover, the minimum spend rising from $20,000 to $50,000, refundability limited to companies less than 10 years old, and supporting activities removed from the program entirely so that only core activities qualify.
Two things to be clear about. First, these changes are announced, not legislated, and the detail can shift before they take effect. Second, the start date is 1 July 2028. Nothing about your FY26 or FY27 claim changes.
What it does mean is that the core and supporting distinction matters even more from here. Businesses that have been running claims weighted heavily toward supporting activities will see a real change in their numbers in a couple of years, and it is worth understanding your own split now rather than being surprised later.
The dates that matter for FY26
Registration with AusIndustry has to be lodged within 10 months of the end of your income year. For a standard 30 June year end, your FY26 activities must be registered by 30 April 2027.
That deadline is absolute. There are no extensions, and a year you miss is a year you cannot go back and claim. Registration also has to happen before you lodge the R&D schedule with your company tax return, so leaving it to the last week creates problems beyond the deadline itself.
Getting your split right
Classifying core and supporting activities properly is not about clever wording. It is about looking honestly at a year of work, finding the points where the outcome was genuinely uncertain, and describing them in the language the program uses.
At Granton we have unlocked more than $7.9 million in R&D refunds across 55 plus clients, with a 100% success rate and zero audits. We work on a success fee, so if a claim does not go through you pay nothing, which also means we only take on claims we are confident in. If your work does not stack up, we will tell you on the first call rather than after you have paid us.
If you are not sure how your year splits between core and supporting activities, book a 15 minute eligibility chat at granton.io/meet. Most founders come off that call knowing exactly which of their activities are claimable and roughly what the claim is worth. You can read more about how the program works at granton.io/rdincentive.
Are you ready to turn your funding aspirations into reality? At Granton, we specialize in helping individuals and businesses navigate the world of grants, offering expert guidance on grant applications and finding opportunities that best suit their needs. Whether you’re seeking funding for a startup, nonprofit, or a specific project, our team is here to assist you every step of the way. We take the guesswork out of Grant Applications, R&D Tax Incentives, and Accelerator Programs, making the process smoother and increasing your chances of success. Ready to take the next step? Book a free consultation with us today, and let’s explore how we can help you secure the grants you deserve. Visit our website at granton.io to learn more or use our contact form to get in touch. Your grant journey starts here!