Eligibility - 5 min read
Is your software company eligible for the R&D Tax Incentive? A 5-minute self-check
A plain-English self-check for Australian software companies. Four questions to see if your engineering work could qualify for the R&D Tax Incentive, what counts as real R&D, and roughly what it could be worth.
If you run an Australian software company, there is a decent chance some of the work your engineers do every week could qualify for the R&D Tax Incentive (R&DTI). The incentive exists to reward exactly the kind of hard technical problem-solving that good software teams do all the time: building something where the answer was not known up front, and you had to experiment to find it.
The catch is that "R&D" in the tax sense is narrower than "engineering". Plenty of valuable, difficult work does not qualify, and claiming work that does not hold up is the dangerous mistake, not the safe one. So before you spend any real time on this, here is a five-minute self-check to see whether you are in the right ballpark.
This is a rough self-qualification guide, not tax advice. The detailed substantiation, and the decision to lodge, is for your registered tax adviser.
The four-question self-check
Answer these honestly. If you can say yes to all four, you are a strong candidate.
1. Are you an Australian company?
You need to be incorporated in Australia (a Pty Ltd), or a foreign company with an Australian permanent establishment, and you pay Australian income tax.
Sole traders, partnerships and trusts do not qualify for the incentive itself. If that is you, it is worth talking to an adviser about your structure, but the self-check ends here.
2. Do you build software in-house?
You have your own engineering team writing code, and that work lives in a place with a real history, like GitHub. The evidence for an R&D claim is in your day-to-day development trail: the pull requests, the commits, the times you reworked something three times before it held. If your software is entirely outsourced or off-the-shelf, there is usually nothing to substantiate.
3. Are you solving problems where the outcome was not obvious?
This is the one that actually decides most claims. The incentive is for technical uncertainty: building something where a competent engineer in your field could not simply look up the answer, buy it off the shelf, or know in advance that the approach would work. You had to form a hypothesis, try it, observe what happened, and adjust.
That is different from commercial uncertainty (will customers buy it?) and different from routine work (standard features, bug fixes, configuration, wiring together existing tools). Hard and time-consuming is not the same as technically uncertain. The tell is something like: "we were not sure it could be done the way we needed, so we had to try a few things to find out."
4. Do you spend at least $20,000 a year on this work?
There is a $20,000 minimum on eligible R&D spend. For most software companies that spend is mostly engineering salaries for the time your team spends on the genuinely uncertain work, not the routine build.
Scoring
- Four out of four: you are a great fit. The next step is to turn that into a substantiated, evidence-backed claim.
- Two or three: worth a conversation. There is often a real claim hiding in here once the eligible work is separated from the routine work.
- Zero or one: probably not a fit right now, and that is a perfectly good answer to have in five minutes instead of five weeks.
What good-fit work actually looks like
To make this concrete, here are the kinds of teams we see qualify most often:
- SaaS and platform companies building genuinely novel features, not just standard CRUD apps.
- Teams doing AI/ML, data engineering, or algorithm work that pushes past what existing tools and libraries can do.
- Work on performance, scale, or reliability where the right approach was not a known quantity going in.
- Hardware, firmware, IoT, fintech infrastructure, biotech tooling, and other deep-tech where the engineering risk is real.
And the work that usually does not qualify:
- Standard websites and brochureware.
- Configuring or customising an existing commercial platform.
- Routine integrations, maintenance, testing and debugging.
- Reproducing functionality that already exists and is well understood.
- Software built purely for your own internal administration.
Roughly what is it worth?
If you qualify, the size of the benefit depends on your turnover.
For companies under $20M in turnover, the offset is refundable. At the 25% base company tax rate that works out to about 43.5% of your eligible R&D spend, and because it is refundable you can receive it as cash even in a year where you made a loss. For a pre-profit startup, that is real money you would not otherwise get.
A rough worked example: $300,000 of eligible R&D spend, under $20M turnover, is in the order of a $130,500 offset.
For companies at $20M turnover and above, the offset is non-refundable: it reduces the tax you owe (at your company tax rate plus an intensity-based premium) and can be carried forward rather than paid out as cash.
These are rough estimates on an FY2026 basis to help you self-qualify, not a calculation of your actual claim. Your real number depends on which activities are eligible and your adviser's review.
What happens next
A self-check tells you whether it is worth pursuing. It does not produce a claim. The work that follows is separating your eligible experiments from your routine build, substantiating each one against the statutory test, linking it to the evidence in your codebase, and estimating the hours and cost. That is the part worth getting right, because a claim that cannot be defended is worse than no claim at all.
If you answered yes to all four questions, that is your green light to take the next step. We prepare the substantiated package; your adviser reviews and lodges it.
This article is general information for self-qualification, 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.