Red flags that put an R&D claim at risk

The R&D Tax Incentive is one of the most valuable programs available to Australian businesses that invest in new or improved products, processes, or software. But it is also one of the most scrutinised. AusIndustry and the ATO jointly administer the program, and both agencies are sharpening their focus on claims that fall short of the legislative requirements.

If you are claiming, or thinking about claiming, you need to know the red flags that put an R&D claim at risk. These are the patterns and gaps that can cause a claim to be pulled for review, delayed, adjusted downwards, or disallowed. And because the incentive is a self-assessed entitlement under the Income Tax Assessment Act 1997, the responsibility for getting it right sits with you, the company, even if you use an advisor.

This guide is not tax, financial, or legal advice. It is general information drawn from published ATO and AusIndustry guidance, and from the experience of registered tax agents who review claims every day. Always confirm your position with a registered tax agent who knows your business before lodging.

Write down the date, too. The rules and rates change. The 2026 reform that proposes lifting the refundable-offset turnover threshold from $20 million to $50 million is just that, proposed, so check the current law for your income year.

With that said, let’s walk through the most common red flags, step by step, and talk about how you can address each one before your accountant presses send.

Prerequisites: what you need before you start

Before you can spot the red flags, you need three things in place.

1. A basic understanding of the R&D Tax Incentive rules. If you are new to the program, start with our plain‑English guide. You need to know what a core R&D activity is, what a supporting R&D activity is, and the difference between eligible and ineligible expenditure.
2. Access to the business’s accounting records, project documents, emails, timesheets, and any other evidence that shows what your team did, when they did it, and how it connects to the R&D activities you plan to claim.
3. A clear list of the projects you believe are R&D. For each project, describe the hypothesis you were testing, the technical uncertainty you faced, the experiments or systematic work you undertook, and the outcome, whether it succeeded or not.

If any of those pieces are missing, you are already looking at red flag number one. Let’s now step through the specific warning signs, grouped by category, along with actions you can take to reduce the risk.

Step 1: Test every activity against the hypothesis requirement

The ATO and AusIndustry are explicit: for an activity to be a core R&D activity, it must be conducted for the purpose of generating new knowledge. That means you need a hypothesis. A belief that something could be achieved, but a genuine technical uncertainty about how to achieve it, or even whether it could be achieved at all.

What a missing or weak hypothesis looks like

Red flag: You describe what you built, not what you were trying to prove.

If your claim narrative simply lists tasks like “developed a new dashboard”, “integrated payment systems”, or “improved machine uptime”, without stating the specific technical problem and the hypothesis you set out to test, it will catch a reviewer’s eye. The ATO’s guidance on Record keeping for the R&D Tax Incentive (ato.gov.au) makes clear that records should show the activities and the purpose. Industry Innovation and Science Australia (IISA), the independent board that advises the government on the program, has also stressed the centrality of the hypothesis when assessing eligibility.

A classic weak description reads: “We built a new mobile app that uses AI to recommend products.” A strong, hypothesis-driven description reads: “We did not know whether we could train a machine‑learning model on our small, noisy transaction dataset to deliver recommendations with an accuracy above 80%. We hypothesised that a few‑shot transfer learning approach using a pre‑trained language model might achieve that, so we designed a series of experiments to test different model architectures and fine‑tuning strategies.”

This is not about fancy words. It is about showing the systematic progression from uncertainty to attempted resolution. That is the backbone of a defensible claim.

How to fix it

Go back to your project team and ask: What did we not know at the start? What was the specific technical barrier? Did we have a written plan or even an email trail that recorded our hypothesis? If not, you can still reconstruct it from contemporaneous records such as sprint goals, design docs, lab notes, or meeting minutes.

At GrantsMAX, the Eligibility Assessment & Risk Flags feature tests each activity your business performed against program rules. It uses your Xero, MYOB, or QuickBooks data plus your project descriptions to flag exactly this kind of gap, before your accountant gets involved. Then the AI Application Pack Drafting turns those validated activities into a narrative that structures the hypothesis and experiments clearly. Remember, the final review and any refinement is done by your registered tax agent through the Accountant Review & Lodge Workflow, who remains in control and lodges on your behalf.

Step 2: Close the record‑keeping gaps before they sink you

Every year, claims are denied or reduced because the business cannot produce evidence that the activity happened and that the expenditure was incurred on that activity. The law requires you to keep records that explain the nature of the R&D activities and the expenditure. The ATO’s R&D Tax Incentive records fact sheet (ato.gov.au) spells this out in blunt terms: you must be able to demonstrate the connection between the costs and the eligible activities.

The contemporaneous record problem

Red flag: The only evidence you have was created after the fact, specifically for the claim.

A project plan written the week before lodgement, timesheets reconstructed from calendar memory, or an email summarising “what we did last year” will not carry the same weight as records that were created when the work was happening. HMRC in the UK, which operates a similar R&D relief scheme, has publicly highlighted that retro‑created evidence is a common trigger for enquiry. In Australia, the ATO and AusIndustry apply the same common‑sense test. For an Australian perspective, CPA Australia’s resource on the Research and Development Tax Incentive also underscores the importance of maintaining records as you go.

How to fix it

If you have gaps, start filling them now for the current year. For claims already in the pipeline, work with your accountant to build a supporting‑evidence index that ties each cost line to the best available document, even if it is not perfect.

GrantsMAX was built to tackle this exact problem. It connects to your accounting system (read‑only) and indexes emails, invoices, and timesheets so that every expense line in the application pack points back to a source. The Audit‑Ready Evidence Trail feature does not create new evidence; it organises what you already have so that your accountant can see the trail clearly. And because the Annual Refresh & Accountant Channel repeats the process each financial year, you stay audit‑ready by default.

Step 3: Purge blanket claims and generic descriptions

A single claim that covers “all R&D” or “ongoing product development” is a magnet for review. The ATO and AusIndustry are not just looking at whether R&D occurred; they are looking at whether each individual activity within a project qualifies.

Why blanket claims trigger review

Red flag: You have one registration for “platform development” that covers everything from DevOps to UI tweaks.

An AusIndustry registration must identify the core R&D activities. If your registration is vague, an examiner has no way to confirm that each element meets the definition in section 355-25 of the ITAA 1997. The same principle applies to expenditure. A single salary allocation of “80% R&D” without any breakdown across projects is often challenged. The UK’s Tax Adviser magazine discussed similar red flags in a recent article, How to avoid R&D enquiries: watch for the red flags, noting that overstated staffing apportionments and vague project descriptions consistently attract HMRC attention. Australian reviewers follow a comparable review logic.

How to replace blanket statements with specific narratives

Divide your claim into discrete projects. For each project, list the core R&D activities separately, and for each activity, describe:

• The technical uncertainty you faced.
• The hypothesis or approach you tested.
• The systematic work you conducted (experiments, prototypes, simulations).
• The outcome, including failures, which are still eligible if the activity met the definition.

Allocate expenditure at the activity level. If an engineer spent time on a non‑R&D task such as routine bug fixing, exclude that portion. Your accounting data should support the split. If you use Xero, you can set up tracking categories to capture R&D project time, and GrantsMAX can then pull that data and map it to the activities you have defined. The Grant & R&D Discovery and Matching engine helps you break down your work into compliant activities from the start.

Step 4: Watch out for misapplied supporting activities

Supporting R&D activities are those that have a direct, close, and relatively proximate relationship to the core R&D activities. They are not a catch‑all.

Red flag: You have claimed large amounts of general “overheads”, “project management”, or “data cleaning” without demonstrating how they directly supported a core activity.

The law is narrow here. For example, administration and corporate overheads are generally not eligible. Even indirectly related technical work, such as routine testing that does not involve the resolution of technical uncertainty, may not qualify. The ATO’s R&D Tax Incentive-Supporting R&D activities page (ato.gov.au) provides detailed examples and is essential reading.

How to test each supporting activity

Ask yourself: If the core activity did not exist, would this supporting activity still have been required? If the answer is yes, it probably is not a supporting R&D activity. For instance, maintaining a server that happens to host an R&D experiment is likely an indirect cost unless you can show that the server was exclusively required for the experiment and would not have been maintained otherwise.

Your registered tax agent will scrutinise this mapping. GrantsMAX’s Eligibility Assessment & Risk Flags module tests supporting‑activity claims against the legal tests, and the Accountant Review & Lodge Workflow flags any line that looks questionable before lodgement.

Step 5: Do not assume all technical work is R&D

Not everything that is technically challenging is eligible R&D. This is one of the most common misunderstandings.

Red flag: You have claimed activities where the outcome was known, the method was routine, or the only uncertainty was commercial rather than technical.

Building a new website using a standard CMS is not R&D, even if your team found it difficult. Customising an off‑the‑shelf ERP system rarely qualifies. On the other hand, developing a novel algorithm that pushes the boundaries of current computer science may well be. The key is whether the activity involved technical uncertainty that could not be resolved by a competent professional in the field without a systematic course of experimentation.

The IRS, which administers the US research credit, published guidance that resonates here: IRS provides facts for businesses about research credit emphasises that the research must be technological in nature and aimed at eliminating uncertainty. While the US rules differ, the core philosophy overlaps with Australia’s, as both draw from the OECD’s Frascati Manual.

How to separate genuine R&D from business‑as‑usual

For every activity you propose, run this test:

• Technical uncertainty: Was there uncertainty about whether a particular technical objective could be achieved, or how to achieve it, that could not be resolved without experimentation?
• Systematic approach: Did you follow a systematic progression of work, forming and testing hypotheses, analysing results, and modifying your approach?
• Outcome: Did the activity generate new knowledge, regardless of whether the project succeeded commercially?

If you can answer yes to all three, and you have the records to prove it, you are in much better shape. Our R&D eligibility guide walks you through these tests with Australian‑specific examples.

Step 6: Get the legal entity and claim ownership right

The R&D Tax Incentive is claimed by the company that conducts the R&D and incurs the expenditure. If a group structure is involved, or if contractors performed some of the work, the claim must be correctly attributed.

Red flag: A subsidiary does the R&D but the parent lodges the claim, or a contractor performs the work but the hiring entity claims the expenditure without meeting the “R&D activities conducted for the company” rules.

This gets technical fast, but the principle is simple: the entity that bears the financial risk and controls the R&D is generally the eligible entity. AusIndustry’s registration system requires an ABN, and the ATO’s company tax return must match the registered entity. Mistakes at this level can lead to an entire claim being disallowed.

How to verify

Map your R&D projects to legal entities. Check your contractor agreements: are you engaging the contractor to perform R&D on your behalf, or are you simply purchasing a service? The difference matters. Ask your registered tax agent to review the structure before you register. If you are using GrantsMAX, the system reads data from your connected accounting file, which is entity‑specific, so the natural starting point is correct; your accountant will still confirm the legal position during the review.

Step 7: Keep an eye on the export market development grant (EMDG) overlap

Many exporters claim the EMDG alongside the R&D Tax Incentive. While the two programs have different objectives, expenditure must be allocated correctly. You cannot claim the same dollar twice.

Red flag: You have claimed marketing or promotional expenditure under the R&D Tax Incentive, or you have included R&D labour costs in an EMDG claim without clear delineation.

Austrade, which administers EMDG, and the ATO share data. A mismatch can trigger a review of both claims. Make sure your accounting system delineates the expenditure by activity type.

How to separate the streams

Work with your accountant to establish clear cost codes for R&D vs. export promotion. GrantsMAX’s Grant & R&D Discovery and Matching engine identifies which grants and incentives may be relevant to your business, including EMDG, so you can plan the allocation from the start.

Step 8: Do not treat the accountant’s review as a rubber stamp

A registered tax agent has a professional obligation to only lodge claims that are reasonably arguable. But the agent is not expected to independently verify every piece of source data; they rely on the information you provide. If you give them a pack that already contains red flags, they may either ask you to fix it, or, if they lodge it and the claim is later found to be incorrect, you are the one facing the amendment, penalties, or interest.

Red flag: You expect your accountant to “sort it out” without providing the underlying evidence.

How to prepare a pack the accountant can stand behind

Give your accountant a complete, evidence‑backed pack. The Why GrantsMAX page explains the division of responsibility: the AI agent prepares the pack from your data, the accountant reviews, refines, and lodges, and you own the claim. That structure respects the Tax Practitioners Board’s requirements and gives your accountant the raw material they need to form their professional judgment.

For first‑time claimants, the GrantsMAX for first‑time claimants workflow deliberately builds a pack that a reviewer can follow, with an evidence index already in place. For accounting firms that serve multiple clients, the white‑label firm workflow lets you standardise the quality of submissions across your portfolio.

How GrantsMAX hard‑wires safeguards into the pack

Every red flag discussed above, missing hypothesis, thin records, blanket claims, misapplied supporting activities, non‑R&D technical work, entity mismatches, program overlap, and a weak accountant handover, can be spotted and addressed before lodgement. GrantsMAX was built to do that systematically.

• The Eligibility Assessment & Risk Flags engine tests your business data against the eligibility tests and identifies where a reviewer would look hardest.
• The AI Application Pack Drafting turns your Xero data, project descriptions, and supporting records into a structured claim, with narratives that foreground the hypothesis and evidence.
• The Audit‑Ready Evidence Trail ties every line in the pack back to a source document.
• The Accountant Review & Lodge Workflow puts your registered tax agent in control, with a clear view of every activity and cost line.
• And because the Annual Refresh & Accountant Channel repeats the process each year, the discipline of good record‑keeping and hypothesis‑led narratives becomes routine.

None of this removes the need for a human accountant. It simply provides the evidence and structure they need to do their job properly, and it gives you, the business owner, confidence that your claim is as robust as possible.

Keep your claim defensible

The R&D Tax Incentive can be transformative for a growing business. But it is not a rebate for simply doing hard technical work. The red flags that put an R&D claim at risk are real, and they are being actively reviewed by the regulators. By testing your activities against the hypothesis requirement, closing record‑keeping gaps, avoiding blanket claims, correctly identifying core and supporting activities, and respecting the legal entity rules, you dramatically lower the chance of an unwelcome letter.

Key takeaways

1. Hypothesis first: Every core activity must have a clearly stated technical hypothesis and a systematic attempt to resolve it.
2. Records must be contemporaneous: Build an evidence trail as the work happens; retro‑created documents are a red flag.
3. Specificity matters: Break projects into discrete activities, and allocate expenditure at that level.
4. Not all difficult work is R&D: Apply the technical‑uncertainty test rigorously.
5. Entity and program boundaries are hard lines: Get the legal entity right, and do not double‑claim across R&D and EMDG.
6. Accountant review is not a wash: Give your tax agent a complete, evidence‑backed pack so they can form a professional judgement.

If you are preparing for your first claim, or if you suspect your current claim might have some of these flags, don’t wait for the review. Take the time now to tighten the documentation and clarify the narratives.

If you would like to see how GrantsMAX can surface the red flags in your data and help you build a pack your accountant can confidently lodge, join the waitlist today at grantsmax.com.