How much is the R&D Tax Incentive actually worth to a startup? (worked examples)

For a pre-profit software startup, the R&D Tax Incentive (R&DTI) is one of the few sources of cash that does not cost you equity. But "up to 43.5%" on a website does not tell you what it means for your company. This post works through real numbers at a few different stages, shows when the cash actually lands, and is honest about what shrinks the figure in practice.

This is general information to help you plan, not tax advice. Your registered tax adviser works out and lodges your actual claim. All figures are on an FY2026 basis.

The mechanic, in one paragraph

If your company turns over less than A$20M, your R&D offset is refundable at roughly 43.5% of your eligible R&D spend (the 25% base company tax rate plus an 18.5 percentage point premium). Refundable means that if you are running at a loss, which most startups are, the offset comes back to you as cash rather than sitting as a credit you cannot use. That cash-when-you-are-losing-money property is the whole reason founders treat the R&DTI as runway.

The number that drives everything is your eligible R&D spend, which for a software team is mostly the salaries of the people doing the genuinely uncertain work, for the share of their time they spent on it.

Worked example 1: two-founder seed startup

A two-engineer team building a novel product. Both spend most of the year on work where the technical outcome genuinely was not known up front.

• 2 engineers, blended cost A$160,000 each: A$320,000 in salary
• Share of time on eligible R&D: ~70%
• Eligible R&D spend: A$224,000
• Refundable offset at 43.5%: about A$97,000

For a seed-stage company burning A$80-100k a month, that is roughly a month of runway, refunded as cash. It is the difference between making it to the next milestone and not.

Worked example 2: post-seed team of eight

A team of eight engineers. The product is further along, so more of the work is now routine build and maintenance, and a smaller share is genuine R&D.

• 8 engineers, blended cost A$170,000 each: A$1,360,000 in salary
• Share of time on eligible R&D: ~40%
• Eligible R&D spend: A$544,000
• Refundable offset at 43.5%: about A$236,000

Notice what changed. The team is four times bigger than example 1, but because a smaller proportion of the work is genuinely uncertain, the eligible base does not scale one-for-one with headcount. This is the normal and correct pattern: as a product matures, the share of true R&D usually falls, even as total engineering spend rises.

Worked example 3: the tempting mistake

Same team of eight, but this time the company claims 80% of all engineering time as R&D, because the work was hard and felt innovative.

• Eligible R&D spend claimed: A$1,088,000
• Refundable offset at 43.5%: about A$473,000

On paper that is twice the example 2 figure. In practice it is a problem. Most of that extra spend is routine development, debugging, maintenance and configuration, none of which is eligible, and software claims are among the most heavily reviewed the ATO runs. An inflated claim is not a bigger benefit; it is an indefensible one that can be denied, with penalties and interest on top. The defensible A$236,000 is worth more than the indefensible A$473,000.

The rate is fixed. The honest size of your eligible base is where the real number lives, in both directions.

When does the cash actually arrive?

This trips up a lot of founders planning runway.

• You register and claim after the income year ends. Most companies have a 30 June year-end, and registration is due within 10 months, so by around 30 April of the following year.
• The refund flows through your company tax return after that, typically a few weeks to a couple of months once lodged.

So R&D spend you incur this financial year generally becomes cash in your account sometime in the second half of the following calendar year. It is runway, but it is delayed runway, which is exactly why some startups use R&D refund financiers to bring it forward. Plan around the timing, not just the amount.

What shrinks the number in practice

The headline figure on a slide is almost always higher than the figure that survives. The usual reasons:

• Eligibility. Only genuine experiments count, not the whole engineering team's time. The share-of-time number does most of the work, and it is usually lower than founders first guess.
• Excluded activities. Routine development, testing, debugging, maintenance, configuring an existing platform, and software for your own internal admin are all excluded.
• Apportionment. You can only claim the portion of each person's time actually spent on eligible activities, and you need a reasonable, documented basis for it.
• The headline is a gross rate. The 43.5% replaces the deduction you would have received anyway. For a loss-making startup the refundable cash is still real new money; for a profitable one the net benefit is the premium portion. Know which you are.

So what is it actually worth?

For an early-stage Australian software startup doing genuinely novel work, a realistic, defensible claim is often in the tens of thousands to low hundreds of thousands of dollars a year, paid as cash while you are pre-profit. The exact figure comes down to two things you control: how much of your spend is genuinely eligible, and how well you can substantiate it.

That is the part worth getting right. A rough estimate tells you whether to bother; a substantiated claim, with each experiment tied to the real pull requests behind it, is what your adviser can confidently lodge and defend.

If you want a quick sense of your own number, our estimate tool will give you one in a couple of minutes, and when you are ready we build the substantiated version from your actual GitHub history for your adviser to review and lodge.

This article is general information, not tax advice, and not a determination of eligibility. Figures are rough FY2026 estimates. Speak to a registered tax adviser before lodging an R&D Tax Incentive claim.