AusIndustry does not publish a checklist of required documents. What it expects is contemporaneous evidence that, for each registered core activity, shows the knowledge gap you identified before starting, the hypothesis you wrote down, the experiments you ran, what you observed, and the conclusions you reached.
The practical work is capturing that while the work happens, then mapping it into the registration fields before the deadline, 10 months after year end (30 April for a 30 June year end).
This is a process guide. If you want the underlying theory of what counts as evidence, start with contemporaneous evidence for a defensible R&D claim.
What AusIndustry requires you to document
Your records have to evidence each element of the statutory test in s 355-25(1) ITAA 1997. First, that the outcome could not be known or determined in advance on the basis of current knowledge, information or experience.
Second, that it was determined by a systematic progression of work based on principles of established science, proceeding from hypothesis to experiment, observation and evaluation, and leading to logical conclusions. Third, that the activity was conducted for the purpose of generating new knowledge.
Nothing in the legislation names a document type. That is deliberate: the question is always whether your records prove the elements, not whether you filled in a template.
Two regulators read the same file. AusIndustry (as delegate of Industry Innovation and Science Australia) decides activity eligibility. The ATO decides expenditure, so it needs a nexus between each claimed dollar and a specific registered activity. Keep records for at least five years.
When should you start documenting?
Before the activity starts, not at year end. Two record types must pre-date the experiments, and they are the ones most claims are missing.
The first is your search of existing knowledge: literature, patents, vendor documentation, standards, technical forums, expert opinion. This is what establishes that a competent professional with access to knowledge available anywhere in the world could not have determined the outcome. In Absolute Vision Technologies [2022] AATA 2319 the absence of a literature review meant no knowledge gap was demonstrated at all, and documents prepared later for an internal review were rejected as non-contemporaneous.
The second is the hypothesis. In Active Sports Management (2023) the tribunal treated the contemporaneous formation of a hypothesis as the essential starting point of the activities. Written after the fact, it is an assertion.
Structuring documentation around activities, not projects
Organise your evidence per core activity, because that is the unit AusIndustry assesses. A useful test: if one experiment's result changes the hypothesis or design of the next, they belong to the same core activity. If two lines of experiments do not inform each other, they are separate core activities.
Getting the granularity right
Most applications land at three to seven core activities. Fewer than three usually means the streams are too broad ("the recommendation engine" is a program of work, not a knowledge gap). More than seven usually means each A/B test or pull request has been promoted to its own activity.
Supporting activities
Group supporting work into two to four broad categories per core activity: prior-art or clinical-literature review, environment and infrastructure setup, data acquisition and labelling, evaluation harnesses, analysis and observability, trial protocol and ethics work. Do not create one supporting activity per ticket. The core versus supporting distinction also carries a dominant-purpose requirement for excluded-category work under s 355-25(2), so record why each supporting activity existed.
What a well-documented hypothesis looks like
A defensible hypothesis is specific, testable, and capable of being invalidated as well as validated. In GQHC [2024] AATA 409 vague and commercially framed hypotheses failed, and the tribunal held that observation and evaluation must be evidenced rather than asserted, with adequate documentation forming part of the systematic progression itself.
Weak: "We hypothesised that our new architecture would improve performance and increase retention."
Stronger: "We hypothesised that sharding the event store by tenant and applying incremental materialisation would hold p95 query latency under 400ms at 50,000 concurrent writers, where our benchmarking of the three published approaches degraded past 8,000. Variables: shard key strategy and materialisation interval. Held constant: hardware class, dataset, query mix."
Record it in a note or plan with a date, list the variables you will vary, hold constant and measure, and state what result would disprove the hypothesis.
The evidence trail for each activity
Each element of the test has an evidence type and a moment at which it has to exist.
| Test element | Evidence | Must exist |
|---|---|---|
| Outcome unknown in advance | Prior-art, patent and literature search results, expert statements, background research showing the gap | Before the activity starts |
| New-knowledge purpose | Project plans, meeting minutes or transcripts, expenditure approvals stating the R&D purpose | At planning |
| Hypothesis | Dated note or experimental plan with the hypothesis and its basis | Before experiments |
| Experiment | Test protocols specifying variables varied, held constant and measured, CRO or research contracts | Before or during |
| Observation | Raw data logs, test reports, timestamped repository and issue-tracker entries, run sheets | During the activity |
| Evaluation and conclusions | Analysis notes, test reports with interpretation, negative results, the next hypothesis | During or after each cycle |
| Expenditure nexus | Timesheets and payroll, invoices, cost allocation to named activities, apportionment method | Ongoing |
Most of these records should live where the work lives. Repositories, issue trackers and CI logs are excellent observation evidence because they are timestamped and hard to backdate. What they rarely hold is the pre-start search and the written hypothesis, so those need a deliberate home: a dated document per activity, linked from the first ticket.
Recording uncertainty as it narrows
The uncertainty itself should narrow on the record as the work progresses. At the end of each experimental cycle, write down what the result ruled out, what remains unknown, and what the next hypothesis becomes.
That running trail is how a reviewer sees the uncertainty being resolved rather than merely claimed, and it is the same material the evaluation and conclusions fields ask for at registration.
Software R&D versus clinical R&D
The test is identical, the knowledge base is not. Keep the two stream types separate, because each is judged against a different competent professional and a different body of literature.
For software, the search evidence is prior art in the engineering sense: published benchmarks, library and vendor documentation, papers, and your own baseline measurements showing where known approaches stop working. See software R&D eligibility for how the s 355-25(2)(h) internal administration exclusion and the (g) reproduction exclusion bite.
For clinical or behavioural R&D, the gap sits in the outcome, not the build. Whether an app measurably improves IBS symptom scores can be a genuine unknown even when the software is routine. Evidence there means the clinical literature review, a protocol with a pre-specified endpoint, ethics approvals, and the analysis against the stated hypothesis.
That pre-specified endpoint is also the line between experimentation and excluded market research. A structured study measuring a clinical or behavioural endpoint against a stated hypothesis is not market research, even when it runs on A/B infrastructure. The same experiment measured on conversion, churn or revenue is excluded under s 355-25(2)(a).
What the registration form asks you to describe
The registration is lodged through the current AusIndustry customer portal form and is structured as projects, then core activities, then supporting activities. The guidance on applying sets out the sequence, the portal steps and processing timeframes.
A core activity is not one narrative. It is split into separate fields with their own character minimums, including the sources you investigated, why a competent professional could not have known the outcome, the hypothesis, the experiment, the evaluation, the conclusions, and the new knowledge generated.
Draft each field to its own question. Repeating your product description across every field is the fastest way to attract follow-up questions.
Turning engineering records into registration-ready text
- List the questions your team could not answer at the start of the year. Those are your candidate core activities.
- For each, pull the dated search evidence and the written hypothesis. If either is missing, note it as a gap now rather than papering over it.
- Allocate signals (pull requests, tickets, trial documents) to the activity whose knowledge gap they probe.
- Convert each signal into hours against an employee and a work date. That allocation is the expenditure nexus.
- Draft the fields from the records, quoting your own measurements and dates. Point at evidence, never invent it.
- Have someone technical who did the work review the drafts against the records before lodgement.
Rand runs this loop from your GitHub and Jira history through the year, groups signals into streams, and drafts each registration field from the underlying evidence for your team and an Expert Reviewer to approve.
A worked example of what is at stake
Step 4 is where the money attaches, so here is what one allocation is worth.
A developer on $180,000 including on-costs spends 40% of the year on a registered core activity. That is $72,000 of notional R&D expenditure. For a base-rate entity under $20M aggregated turnover, the refundable offset at 43.5% is $31,320.
If timesheets cannot support the 40%, or the activity is found ineligible for want of a hypothesis, that $31,320 is what is under review.
One threshold worth knowing: claims below $20,000 in notional deductions are only available for amounts paid to a Registered Research Service Provider or as CRC contributions.
Reforms to the R&DTI announced in the 2026-27 Budget apply only to income years starting on or after 1 July 2028. The rates and thresholds above reflect the current rules.
Documentation mistakes that trigger follow-up questions
- A hypothesis that is a commercial goal. GQHC [2024] AATA 409 is the direct authority.
- No documented search of worldwide knowledge, so no demonstrated gap. New to your team is never enough.
- Trial and error without isolable variables. "We tried a few things until it worked" is not a repeatable experiment.
- Observation recorded, evaluation missing. No conclusions, including negative results, means the progression is incomplete.
- Records that do not match the registration. A mismatch between the two was fatal in Absolute Vision.
- No expenditure nexus. In the ATO-cited Tier Toys and Ozone Manufacturing matters, absent contemporaneous records the taxpayer could not show the spend was more than normal business expense.
- Treating novelty of context as automatic eligibility. Moreton Resources [2019] FCAFC 120 allows a new context to support eligibility, but Coal of Queensland [2021] FCAFC 54 confirms the four-part test and its evidence still have to be met.
Making it repeatable next year
Build the habit into the delivery cycle rather than the reporting cycle.
At the start of each activity, create a dated one-page record: the question, the search results, the hypothesis, the variables, the success and failure criteria. During the work, tag tickets and pull requests to the activity and keep test outputs.
At each cycle's end, write a short evaluation note including what failed. Quarterly, reconcile hours to activities and check for streams that have drifted too broad or too narrow.
Do that and the registration becomes a transcription job rather than an archaeology project, and an ATO or AusIndustry review is answered from records you already have. Check the ATO's R&DTI guidance and the business.gov.au program pages for current administrative detail, and confirm your lodgement dates against the deadlines that matter.
Rand exists because this evidence already sits in your engineering and clinical systems, and the compliance work is mostly a matter of organising it while it is still contemporaneous.