Keep records that show the hypothesis, the experiments, the results and the conclusions for each registered activity, and that link expenditure to those activities. That sentence from the jurisdiction record is the whole rule, and it reflects the structure of the law: the definition of a core R&D activity is written as a sequence of steps, so the evidence has to show those steps happening. Two agencies will look. The department examines whether the registered activities were conducted and were eligible; the ATO checks that the expenditure was incurred on them.
What the tax authority expects
Neither agency prescribes a format. The department's guidance asks for records that were created at the time, that show the activities as registered actually took place, and that demonstrate each element of the definition: that the outcome could not be known in advance, that a hypothesis was formed, that experiments were conducted and observed, that results were evaluated and conclusions drawn, and that the purpose was to generate new knowledge. For supporting activities, the records must show the direct relationship to a core activity and, where the dominant purpose test applies, that purpose.
The ATO's interest is the expenditure: that it was incurred, that it was incurred on registered activities, that apportionments were reasonable and documented, and that the associate, feedstock and clawback rules were applied. The ATO and the department have issued taxpayer alerts identifying R&D risk areas.
Records to keep during the year
| Record | Shows | Kept by | When |
|---|---|---|---|
| Literature or state-of-knowledge search | Why the outcome could not be known in advance | Technical lead | Before the experiments start |
| Hypothesis statement | What was proposed and how it would be tested | Technical lead | At the start of each core activity |
| Experiment plans, lab notebooks, test logs, build records, version control | The experiments and observations | Scientists and engineers | As the work happens |
| Results analysis and conclusions | Evaluation and the knowledge generated, including failures | Technical lead | At each milestone and at the end |
| Activity register | Which activities are core, which supporting, and the link between them | Project manager | Updated through the year; becomes the registration |
| Timesheets by activity | Staff time on each registered activity | All R&D staff | Weekly |
| Invoices, contracts and payment records | Contractor and associate costs, dates paid | Finance | At purchase and payment |
| Asset register and usage records | Decline in value and R&D use percentage | Finance | Monthly |
| Grant agreements | Recoupments that trigger clawback | Finance | On signing |
None of these records has to be created for the claim. A lab notebook, a ticketing system, a commit history and an ordinary timesheet serve, provided they can be tied to an activity. A hypothesis written in July beats a narrative written the following April, and both agencies know the difference.
Linking records to projects and costs
The registration is the link. Each registered activity has a name and a number; each timesheet entry, purchase and asset usage record carries the activity name; and the schedule workpapers map each activity to the expenditure categories. A minimum workpaper for one core activity:
| Item | Content |
|---|---|
| Activity | Registered name and number, start and end dates, the knowledge gap, the hypothesis |
| Evidence index | Pointers to the experiment records, results and conclusions |
| Supporting activities | Which ones, and the dominant purpose reasoning where it applies |
| Salary expenditure | Staff, hours or percentage, salary and on-costs, apportionment basis |
| Contractors and associates | Contract, invoice, date paid, whether an associate |
| Materials and feedstock | Purchases, whether transformed into outputs sold, feedstock adjustment |
| Assets and overheads | Decline in value and R&D use; overhead apportionment method |
| Adjustments | Clawback for recoupments, expenditure not at risk |
Common gaps
- Project-level registration. Registering "develop product X" as one core activity, with no hypothesis. Examinations routinely find this ineligible.
- No evidence of the knowledge gap. Nothing shows that a competent professional could not have known the outcome; a short literature search at the start fixes this.
- Narratives without experiments. Descriptions of what was built rather than what was tested and learned.
- Whole-of-staff apportionments. Round percentages of everyone's time, with no timesheets.
- Associate payments claimed when incurred. They are deductible only when paid.
- Grants not adjusted. A government grant for the same expenditure requires a clawback adjustment.
- Overseas work without a finding. Offshore contractors claimed without an Overseas Finding.
Illustrative example
Illustrative example. Warrego Agtech Pty Ltd, a hypothetical company, develops a soil moisture sensor for clay soils. Its agronomist's July note records that published calibration curves fail in high-clay soils and proposes a hypothesis about dual-frequency measurement. The engineers log each field trial with soil type, readings and reference measurements, and tag their firmware commits to activity WA-1. Finance codes sensor components and the university's testing invoice to WA-1, and timesheets show 70% of two engineers' time. When the department examines the registration two years later, Warrego answers from the file in a week and the finding is positive. The same work with a single April narrative and a 70% estimate would be a long argument.
Related pages
Official sources
Educational content only, not tax advice. See the disclaimer.