There are two ways to think about ATO and AusIndustry compliance on an R&D claim. The first is defensive: lodge the claim, wait, hope no questions arrive, and if they do, scramble to assemble the evidence to support what was already submitted.
The second is structural: design the claim so the evidence is in the file the day it's lodged, queries can be answered from records that already exist, and the audit response is a half-day of work rather than a six-week project.
Both approaches result in lodged claims. Only one survives scrutiny without disruption.
What goes wrong with audit defence
The defensive model assumes audit risk is something that happens to other companies. It works until it doesn't, and when it stops working, the costs compound:
- Memory has degraded. The technical lead who could have explained the hypothesis on the day the work was done has changed roles, left the company, or simply forgotten the detail. The reconstruction is approximate.
- Documentation is retrofitted. Project descriptions, experiment logs, and decision rationale get drafted from outcomes rather than from the actual sequence of work. The narrative reads as plausible but isn't anchored to contemporaneous records.
- Apportionment can't be reconstructed. Time spent on R&D versus routine engineering, contractor allocations, overhead methods — these get estimated rather than derived. Estimates don't hold up under structured ATO inquiry.
- The claim survives or fails on the strength of the response, not the underlying work. Even where the work was genuinely eligible, a poorly-supported response can result in disallowance.
The defensive model also has a cumulative effect. A query in year three may surface inconsistencies between the year-three position and the year-one position, opening earlier years to review.
What audit-ready by construction looks like
The alternative is to build the claim with the eventual ATO question in mind from the start. The components are:
Contemporaneous documentation.
Hypothesis statements, experiment logs, and decision points are captured at the moment the work happens, not at year-end. The format doesn't matter — engineering team retros, design documents, internal wiki pages, source control commits with structured messages, dedicated R&D logs — what matters is that the record is dated, attributable to a person, and predates the lodgement.
Evidence in step with expenditure.
Every dollar of eligible expenditure traces to a record that explains what it was spent on, why it qualifies, and who made the call. Payroll attribution to R&D activity is decided and recorded during the year, not estimated at June 30.
Technical literature scans done up front.
The competent professional test requires that the work wasn't trivially solvable from publicly available knowledge. A literature scan done before the registration is lodged, showing what was looked at and why it didn't answer the question, is materially more defensible than one done after a query lands.
Apportionment methodology written down.
Time allocation rules, contractor allocation rules, overhead methods, all documented and applied consistently. When the ATO asks how 40% of an engineer's time was attributed to R&D, the answer is a methodology, not a calculation done in the moment.
Decision audit trail.
Where eligibility calls were made — including the calls to exclude activities that didn't qualify — the reasoning is on file. Conservatism on individual items reads as discipline rather than oversight.
The operational cost
The structural approach costs more in the early days of an engagement and less every year after. Year one builds the infrastructure: the templates, the conventions, the cadence of capture. Year two runs on the infrastructure and the marginal effort drops materially. Year three and beyond, the documentation is largely a by-product of how the engineering team already works.
The defensive approach costs less in year one and increasingly more in every subsequent year, as the gap between what was lodged and what could be evidenced widens.
The audit itself
When an ATO or AusIndustry query lands against a claim built audit-ready, the response cycle looks like:
- Pull the relevant project description from the file.
- Pull the corresponding experiment logs, hypothesis statements and decision points.
- Pull the apportionment methodology and the supporting expenditure traces.
- Draft a response that quotes the contemporaneous records.
- Lodge.
The total elapsed time is usually a half-day. The response doesn't ask the team to remember anything, doesn't require fresh evidence collection, and doesn't put the underlying work at risk. The claim was built to be defended; defending it is the predictable outcome of how it was built.
That's the posture. Not audit defence. Audit posture.