An accelerator sits with its companies at exactly the moment the R&D Tax Incentive is easiest to get right: before the records exist. Twelve weeks of guidance on records and ownership can be worth more to a cohort company's first claim than anything done at year end. The claim is normally the cohort company's. An accelerator entity that conducts eligible R&D of its own could claim for that like any other company, but mentoring, selection and running the programme are not R&D. This page is for the programme team, not the founders; founders have their own page.
What a cohort company needs in place
The incentive belongs to a company, so the first conditions are structural. The team must be incorporated, and only an R&D entity can register. The R&D must be conducted for that company: it directs the work, carries the cost whatever the outcome, and effectively owns the results. Founders assigning their code and IP to the company is the usual way to make that clear. It must spend at least $20,000 on eligible R&D in the year, unless the work was paid to a registered research service provider that is not an associate or the spend is a Cooperative Research Centre contribution. And its records must show the work as it happened: AusIndustry asks for background research, what was done, when and by whom, and the expenditure tied to each activity.
For a cohort company with a 30 June year end, a programme running from February to May sits inside one income year. Registration for that year opens on 1 July and closes on 30 April. The registration field checker checks drafted answers against the form's limits; it is not an eligibility test.
Programme work that is not R&D
Most of what an accelerator schedules is valuable and is not R&D:
- pitch coaching and demo day preparation;
- customer discovery interviews, surveys and market validation (market research is excluded);
- business model and pricing experiments;
- building a demo to show investors when the technology is already known to work.
R&D is narrower. It is an experiment whose outcome a competent professional could not know in advance, run from a hypothesis through observation to a conclusion. Early prototypes often qualify only in part; MVPs and prototypes explains where the line falls.
Funding and ownership traps
University programmes. AusIndustry's form asks whether income tax exempt entities and their affiliates hold at least 50% of the voting power or of rights to income or capital. A spin-out a university controls cannot have the refundable offset.
Grants. A government grant or other recoupment for the same R&D usually triggers a clawback adjustment, which adds back part of the benefit. Some funding works differently: money received whatever the results, for example from a customer or programme partner, can make the expenditure not at risk under s 355-405, which takes it out of the claim. Teams combining funding with the incentive should have the accountant check which rule applies before forecasting both.
Venture studios. A studio that controls several companies may make them connected entities, whose turnover is then added together for the $20 million test. A studio that keeps the IP and licenses it to each company raises the separate question of whether the R&D is really conducted for the company that claims it. Both are worth settling with an accountant before the first claim.
A week-one checklist for cohort companies
- Confirm the company is incorporated and registered with ASIC.
- Assign any founder-held code, designs or IP to the company.
- Name the technical unknowns you expect to test this year.
- Write each hypothesis down before the first experiment.
- Record results and conclusions, including failures, as they happen.
- Keep payroll, invoices and time in a form you can tie to each activity, and keep it for five years after the claim is made.
Strong habits here also make the contemporaneous evidence test straightforward, and keep a company's options open if it later needs an advance finding.