Australia’s R&D Tax Incentive in 2026: current rules, proposed reforms and the records to prepare

Australia’s R&D Tax Incentive is at a useful planning moment. Companies have current claims to prepare while the government is proposing a different program for future income years. Treating those two timelines as one can distort both a claim and a budget.

As of 7 October 2026, the Department of Industry, Science and Resources says it will continue administering the program under current legislation until the announced changes for income years starting on or after 1 July 2028. A budget announcement and an exposure draft should be tracked as proposed policy when planning; they do not replace the rules for a current claim. DISR program and reform update.

For a finance or engineering leader, the immediate opportunity is to make the company’s actual research legible: what it tried to discover, who performed the work, what it cost and which facts a reviewer can verify.

The current rate depends on the company

Under the current settings, eligible entities with aggregated turnover below A$20 million generally receive a refundable offset at their company tax rate plus 18.5 percentage points, subject to the exempt-entity control rules. Other eligible entities receive a non-refundable offset: the company tax rate plus 8.5 points on the expenditure within the first 2% of total expenses, and plus 16.5 points above that intensity threshold. The premium stops applying to notional deductions above A$150 million. The usual A$20,000 minimum has specified research-service-provider and cooperative-research exceptions. These are current statutory settings, not a rate applicable to every software company. Income Tax Assessment Act 1997, section 355-100.

Illustrative arithmetic: if an eligible refundable-offset company has a 25% tax rate and A$200,000 of qualifying notional deductions, 43.5% produces an A$87,000 gross offset. That figure is not automatically cash received, incremental profit or project ROI. The company’s tax position and the treatment of the expenditure still matter. Keep the gross offset separate from the adviser’s assessment of the net tax effect.

Our practical recommendation is to put three columns in the management model: engineering expenditure, expenditure proposed for tax review, and expenditure accepted after review. Keep rejected and unresolved amounts visible. Applying a headline percentage to the first column obscures the decisions that matter most.

What the 2028 exposure draft proposes

Treasury consulted on draft reforms from 10 to 28 September 2026. Its summary proposes increasing support for core R&D while removing supporting-activity eligibility; lowering the non-refundable intensity threshold to 1.5%; increasing the refundable turnover threshold to A$50 million; increasing minimum and maximum expenditure thresholds to A$50,000 and A$200 million; and limiting refundability by company age, with a longer period for specified therapeutic-goods research. These proposals target commencement from 1 July 2028. This article treats them as draft measures, not enacted current settings. Treasury exposure-draft consultation.

That proposed shift makes the separation of core and supporting activities particularly useful for planning. A company could model how its spending mix would be affected if the draft were enacted, while retaining its current-law assessment for the claim being prepared. Label that scenario with the draft date and assumptions. Do not rewrite historical activity descriptions to make the future scenario look better.

For example, a technical team may track experimental model development, experiment-specific test infrastructure and ordinary deployment work as separate activities. Even before a tax decision, that distinction helps management understand how much it spent discovering a solution and how much it spent putting a known solution into service.

Registration is a deadline, not a verdict

The ordinary registration deadline is 10 months after the company’s income-year end. For a 30 June 2026 year-end, that means 30 April 2027. The government expressly warns that confirmation of registration does not establish activity eligibility; the program relies on self-assessment. Use the company’s actual income year, particularly where a substituted accounting period applies. Application requirements and registration guidance.

Build the operational calendar backwards from the filing process. Leave time for a technical interview, financial reconciliation, an unresolved-questions pass and a responsible person’s review. A sensible internal target is a complete evidence inventory well before the statutory deadline. That is a workflow choice, not another legal deadline.

The inventory should distinguish “requested,” “received” and “reviewed.” A connected source can still have missing months, inaccessible repositories or records that do not address the question. A full upload folder can still lack an explanation of the experiment.

Prepare two linked views of the same work

Start with an activity view. Give each candidate activity a stable reference and record the technical objective, the uncertainty at the time, the approach tested, dated observations and the conclusion. Link exact source records, including results that caused the team to abandon an approach.

Then construct a cost view. Preserve ledger references, payroll periods, supplier invoices, credits, currencies and any proposed allocation method. Link a cost to the activity it concerns and identify the person who can explain the relationship. A project name typed into a billing spreadsheet is a lead for investigation, not a completed allocation.

The department’s record-keeping guidance calls for records created while the work takes place and links between activities and expenditure. Our proposed two-view structure is one practical way to organize that material; the government does not require this particular software or template. R&DTI record-keeping guidance.

The ATO and DISR have different administrative responsibilities: activity registration and assessment sit with DISR, while the ATO deals with the allowable expenditure and tax offset. A coherent engineering narrative therefore cannot substitute for checking the money. DISR program responsibilities.

Use AI to shorten investigation without inventing a history

An AI-assisted workflow can be useful for finding repeated project names, suggesting which records describe the same experiment and drafting questions for an engineer. Design it to show the source behind each factual sentence. Ask it to preserve contradictions: an issue may say the experiment succeeded while the evaluation report shows an unresolved failure.

Keep later explanations visibly separate from contemporaneous records. A summary written in October should retain its October creation date even when it describes a March experiment. Ask the engineer to verify the facts, then let the responsible adviser assess the treatment.

Proforma’s starting offer is an initial evidence review: supplied records, supported project notes, available costs, gaps and next questions. Preparation can be manual. It gives a company and its adviser a concrete starting point for the work above; signing up does not produce an eligibility determination or a lodged claim.

Next step: bring one project’s technical records and cost exports to an initial review. A small, inspectable case will reveal more about readiness than a company-wide percentage estimate.