Most companies that claim the R&D Tax Incentive treat it as an annual event. The year ends, someone pulls together what was done, a registration is prepared, a schedule is lodged, and the work is shelved until the following April.

This works. It also leaves a lot on the table.

Why the annual model under-delivers

A single-year claim treats each twelve-month period as a discrete event. That framing forces three compromises:

  1. The evidence is always retrofitted. Project descriptions are written months after the work happened, when the team's memory of the hypothesis, the experiments and the technical uncertainty has degraded. The narrative reads as a tidy story rather than a record of how the work actually unfolded.
  2. Multi-year research questions get fragmented. Genuine R&D often runs over several years. A hypothesis tested in FY24 might not resolve until FY26. Reframing the question annually loses the thread, and the second and third years of work get described as new projects rather than continuations.
  3. Programme-level efficiencies aren't captured. Apportionment methodologies, expenditure categorisation logic, supporting evidence templates — these are all built fresh each year. The compounding benefit of refining them never materialises.

What a programme view changes

A multi-year R&D programme treats the claim as continuous. Activities are registered each year, but the underlying programme has a defined scope, articulated hypotheses, a planned trajectory, and a documentation infrastructure that accumulates rather than restarts.

Operationally, the changes are practical:

  • Documentation runs alongside the work, not after it. Technical leads capture experiment logs, design rationale and decision points as they happen. By year-end, the project description writes itself.
  • Apportionment methodology stabilises. Time tracking conventions, contractor allocation rules and overhead methods are agreed once and refined incrementally. Year-on-year variance becomes interpretable rather than arbitrary.
  • Evidence templates compound. A library of technical literature citations, hypothesis statements and outcome records builds across years. Year three's documentation effort is materially lower than year one's.
  • Audit responses become predictable. Because the work is documented in step, ATO and AusIndustry queries can usually be answered from records that already exist in the file. The cost of a query is hours rather than weeks.

When the programme view makes sense

A programme view isn't right for everyone. The threshold question is whether the underlying work is continuous. A one-off project with a defined endpoint may not justify the operating overhead. A first-time claimant may need a single-year claim first, to learn the structure before committing to a programme.

The clearest signal is the shape of the company's research questions. If the questions span income years — if you're still testing the same hypothesis at the end of June that you were testing the previous July — the work is programmatic in nature whether you treat it that way administratively or not.

The compounding effect

In year one, a programme runs at roughly the same effort as a well-structured annual claim. In year two, the documentation infrastructure starts paying back. By year three, the programme is materially less effort to run than the annual equivalent, more defensible against ATO scrutiny, and easier to use as the basis for forward-looking decisions about which R&D to prioritise.

Most of the established R&D-active companies we work with run a rolling programme with annual lodgements. The administrative load is lower, the claims are stronger, and the firm has a usable record of its own research trajectory at the end of every income year.