Every R&D Tax Incentive claim is built from two kinds of activity, and they are not equal. Getting the distinction right has always mattered for eligibility. After the 2026-27 Budget, it becomes one of the most consequential decisions in your claim, because from 1 July 2028 supporting activities are proposed to lose eligibility entirely.
This article explains the difference under Division 355 of the ITAA 1997, how to classify your work, and why the classification is about to carry far more weight.
The two categories
Core R&D activities are the experimental work itself. To be core, an activity must meet all four criteria: it is experimental, its outcome cannot be known in advance (tested against the competent-professional standard), it follows a systematic progression from hypothesis to conclusion, and its purpose is to generate new knowledge.
Supporting R&D activities are activities directly related to a core activity. They are not experimental in their own right, but they enable, feed, or follow from the experiment. Where a supporting activity also produces goods or services, or is an excluded activity, it must additionally pass a dominant purpose test, its dominant purpose has to be supporting the core R&D.
The relationship is directional
A supporting activity must point at a specific core activity. You cannot have supporting activities floating free. If you cannot name the core activity a piece of work supports, it is not a supporting R&D activity, it is just work.
How it looks in software
| Activity | Classification | Why |
|---|---|---|
| Designing and running an experiment to resolve a genuine technical unknown | Core | It is the experiment |
| Building test harnesses or scaffolding specifically to run that experiment | Supporting | Directly enables the core activity |
| Preparing or cleaning a dataset solely to test the hypothesis | Supporting | Feeds the experiment; check dominant purpose |
| Routine feature development with known methods | Neither | Not experimental, supports no core activity |
| Production deployment and BAU maintenance | Neither | Operational, not R&D |
| Requirements gathering and project management | Usually neither | General business activity |
For the full eligibility test on the core side, see the software eligibility guide.
The classification mistakes that cost claims
- Over-claiming supporting. Sweeping broad swathes of normal development into "supporting" because it happened near the experiment. Proximity is not direct relation, and the dominant purpose test catches a lot of it.
- Mislabelling core as supporting (and vice versa). Some genuinely experimental work gets filed as supporting out of caution, understating the core. The reverse, dressing routine work as core, fails on the four criteria.
- Orphan supporting activities. Supporting work with no clearly named core activity behind it.
A simple test
For each activity, ask two questions. Is this the experiment, or directly tied to a specific experiment? And would I be doing it, at this scope, if the R&D were not happening? Core is the experiment. Supporting is tied to it and would not exist at this scope without it. Everything else is neither.
Why 2028 changes the stakes
Today, both categories are eligible (supporting subject to the dominant purpose test). The 2026-27 Budget proposes that from 1 July 2028, the offset applies to core activities only, with higher rates to partly compensate. The detail and full set of changes are in the 2026-27 Budget explainer.
The practical consequence: the line between core and supporting stops being an accounting detail and becomes the line between claimable and not. Two things follow.
- Classification accuracy becomes financial. Misfiling core work as supporting will, from 2028, mean leaving eligible benefit on the table. Misfiling routine work as core will mean an indefensible claim. Precision pays both ways.
- Some work deserves a fresh look. Work previously filed as supporting may, when examined against the four criteria, genuinely belong in core. This is not relabelling for advantage, it is describing the work accurately. Done honestly, it keeps eligible work eligible.
What to do now
You have financial years before the change lands. Use them to audit your core-to-supporting split, model your claim with supporting removed and the higher core rate applied, and tighten how you describe the experimental core. Good contemporaneous evidence underpins all of it, see documenting a defensible claim.
The R&D Tax Incentive is a self-assessment program. Rand helps you prepare and document a defensible claim, but your company and its directors remain responsible for its accuracy, and you should seek advice specific to your circumstances.