Back to the blog

How to document R&D so it survives an audit

A practical record-keeping playbook for Australian tech companies. The three evidence pillars, the five-step scientific method, and the audit-proof checklist — without bolting on a separate documentation system.

5 yrs
record retention (s 262A ITAA 1936)
4-part
core R&D test your records must satisfy
3 pillars
activity · expenditure · linkage evidence
Published 9 May 2026 · ClaimKit
1Why documentation isn't optional

The R&DTI is self-assessed — your records are the claim

The R&DTI is a self-assessment program. You — the company — decide whether your activities and expenditure are eligible. The ATO and DISR don't pre-approve claims. They review them after the fact. If a reviewer asks you to demonstrate eligibility and you cannot produce contemporaneous evidence, you are not entitled to the offset — regardless of whether the R&D actually happened.

The DISR Guide to Interpretation puts it bluntly: a taxpayer cannot succeed in establishing eligibility in the absence of detailed documentation recording the process of each activity as it develops. Documents created after the fact will generally not be adequate on their own.

The cost of weak documentation is concrete: full or partial disallowance of the R&D tax offset, ATO interest and penalties on amended assessments, registration cancellation by DISR, and reputational exposure if findings are published.

Records have to track to the four-part test

Every piece of documentation you create should be traceable back to the four-part core R&D test in s 355–25(1) of the ITAA 1997. If your records can't speak to each element, your claim is at risk. Each element needs its own type of evidence:

Element
What your records must show
(a) Unknown outcome
s 355–25(1)(a)
Background research showing current knowledge was insufficient. Expert opinions, literature searches, patent searches, technology reviews.
(b) Systematic progression
s 355–25(1)(b)
Hypothesis statements, experiment plans, test logs, results records, analysis write-ups, revision notes.
(c) New knowledge
s 355–25(1)(c)
Project plans, board minutes, funding approvals, email correspondence stating the purpose of the activity.
(d) Not excluded
s 355–25(2)
Analysis of dominant purpose of software use. Evidence the software is used by external customers — not for your own admin.
Software companies — read this carefully. The internal administration software exclusion is the single biggest compliance risk you face. If the software you're developing is for your own internal use (payroll, ERP, CRM, invoicing), it's excluded from core R&D. Document the dominant purpose of the software before you claim.

Two dates that matter

  • Register within 10 months of your income year end (e.g. 30 April for a 30 June year-end). Registration is not approval — it's a declaration that you have the records to support a claim.
  • Retain records for 5 years after the income year in which you claimed. Activity evidence, expenditure evidence, linkage evidence, and correspondence with DISR or the ATO all fall under record retention.
Set up your system now. Build documentation into how you already work — sprint planning, code reviews, retros — before you start R&D activities. Not at year-end when you're writing the application.

2The evidence framework

Three pillars and five steps

Your documentation must cover three areas at the same time. Think of them as three separate files you maintain throughout the year. Each pillar answers a different question a reviewer will ask.

1
Activity evidence
That your activities are eligible R&D — core or supporting.
  • Hypothesis documents
  • Experiment plans & protocols
  • Test logs and results
  • Evaluation notes
  • Conclusion records
2
Expenditure evidence
What you spent and on what — wages, contractors, assets, overheads.
  • Payroll records & timesheets
  • Contractor invoices & contracts
  • Purchase orders
  • Asset register & depreciation schedules
  • Overhead invoices
3
Linkage evidence
That the expenditure was incurred on the eligible R&D activity.
  • Time-allocation records
  • Apportionment methodology
  • Project codes in financial system
  • Activity-to-cost mapping
  • Sprint / ticket linkage
All three pillars must be maintained. A claim that is strong on activity evidence but weak on linkage will still be disallowed on expenditure.

The five-step scientific method

For each core R&D activity, your documentation must walk through every stage of the systematic progression of work. Click each step to see what to record — and a template you can use.

Your hypothesis is your proposed explanation for how you could achieve a particular technical result, and why. It must be developed before you begin experimental work.

  • What technical result you are trying to achieve
  • How and why you believe it may be achievable, based on your background knowledge
  • What is currently unknown that prevents you from achieving it without experiment
  • The date the hypothesis was formed and who prepared it
Falsifiable-statement template
We hypothesise that [approach X] will achieve [technical outcome Y] because [scientific rationale Z]. This cannot be confirmed without experimental testing because [gap in current knowledge].
Authority: Guide to Interpretation, May 2024 §4.1

Supporting R&D activities — what to add

For supporting activities, your records must additionally show:

  • Direct relationship. How and why the activity is directly connected to a specific core R&D activity. The connection must be more than incidental — it must be essential to conducting the core activity.
  • Dominant purpose (where required). For activities producing goods or services, or activities directly related to producing goods or services: evidence that the primary reason for conducting the activity was to support the core R&D — not to generate commercial output.
  • Timing. Supporting activities can occur in a different income year to the core R&D activity they support. Document the temporal relationship clearly.
Don't over-scope. Marketing, sales, general administration, and routine software maintenance are not supporting R&D. The connection to a specific core activity must be direct and documentable.

3Your existing tools = your evidence

You don't need a separate R&D documentation system

You need your existing systems to capture the right information. The trick is connecting them into a coherent evidence trail — hypothesis in Notion, experiments in Jira, iterations in GitHub, results in test reports, time in your time-tracker. Click any row to see what to capture in each tool.

Build evidence capture into your sprint ceremonies. A brief R&D check-in at retros — what unknowns were resolved, what new unknowns emerged, what was learned — generates valuable contemporaneous documentation with minimal overhead.

Expenditure documentation — the ATO side

The ATO reviews R&D expenditure separately from DISR's activity review. Your records must show three things: that the expenditure was actually incurred, that it was incurred on an eligible R&D activity, and that reasonable apportionment has been applied where expenditure relates to both R&D and non-R&D work.

Wages & salaries
Payroll records, employment contracts, timesheets or time-allocation records against R&D project codes.
Contractor payments
Invoices, contracts describing the R&D work, statements of work, evidence of payment. The at-risk rule applies — the expenditure must be at your financial risk.
Decline in value
Asset register, evidence of R&D use (what asset, when, for which activity), depreciation calculations.
Overheads
Invoices, usage reports, written apportionment methodology — different methods may suit salaries vs utilities vs cloud compute.
Apportionment based on a guess is not apportionment. “We estimated 50% of developer time was R&D” without underlying time records is not acceptable. Reviewers will ask what contemporaneous data the estimate was based on.

Common documentation failures

The same handful of failures account for most disallowances:

Activity scoped as the entire project
Break the project into discrete experimental challenges. Each challenge with an unknown outcome is a separate core R&D activity.
Hypothesis written at year-end to match what was done
Write hypotheses before experiments commence. Use sprint planning or design documents as the starting point. File metadata can betray retrospective writing.
Results captured but evaluation missing
For every experiment, write a brief evaluation: what the results mean, why they occurred, and whether they support the hypothesis.
No staff time records
Implement project-code timesheets. Even retrospective timesheets backed by Jira tickets and commit logs beat nothing.
Software claimed without checking s 355–25(2)(h)
Before including any software activity, document who the software is for and what its dominant purpose of use is.
BAU work bundled into R&D scope
Separate genuine technical unknowns from commercial development. Routine builds, bug fixes, and BAU maintenance create audit exposure across the entire claim.

4Pre-lodgement checklist

Run this before you register

Use the checklist as your annual quality check before lodging your R&DTI application. Tick each item you can produce contemporaneous evidence for. State only — your ticks aren't saved anywhere.

Pre-lodgement readiness
Tick each item you can produce contemporaneous evidence for. Every “no” is a compliance risk.
0/17
0% ready
Activity eligibility
Supporting activities & expenditure
Administration
Not sure how to close the gaps? The line between adequate and audit-proof documentation is often fact-specific. A quick chat with the ClaimKit team can help you work out where to focus first.

5Glossary

Key terms defined

The R&DTI uses specialised legislative language. These definitions refer directly to the legislation and official guidance. The same terms are highlighted inline throughout the article — click any dotted-underlined term to pop up its definition.

ContemporaneousGuide to Interpretation, May 2024
Created at or near the time of the activity it describes — not reconstructed afterward. The DISR Guide to Interpretation is explicit that documents created after the fact are generally inadequate on their own to establish eligibility. File metadata (creation dates, authors) can be used by reviewers to assess whether records are genuinely contemporaneous.
Self-assessment programDivision 355, ITAA 1997
The R&DTI is self-assessed: the company decides whether its activities and expenditure are eligible, registers, and claims the offset. The ATO and DISR review claims after the fact. Registration is not approval — it is a declaration that you have the records to support your claim.
DISRIR&D Act 1986
The Department of Industry, Science and Resources. Administers the activity-side of the R&DTI: registration, advance findings, overseas findings, and reviews of whether your activities are eligible core or supporting R&D. Publishes the Guide to Interpretation, which is the primary reference for activity documentation.
HypothesisGuide to Interpretation, May 2024 p.17
A proposed explanation for how or why a particular technical result might be achievable. Must be developed before experiments begin, must have a scientific basis, and must be testable through experiment. Records should show the hypothesis was established prior to the commencement of experimental activities — retrospective hypothesis writing is explicitly flagged as inadequate.
Systematic progression of works 355–25(1)(a), ITAA 1997
The five-element scientific process required for core R&D activities: hypothesis → experiment → observation → evaluation → logical conclusions. Must be based on principles of established science. Records must show that activities proceed from one element to the next.
Core R&D activitys 355–25(1), ITAA 1997
An experimental activity whose outcome cannot be known in advance based on current worldwide knowledge, that follows a systematic progression of work based on established science, and that is conducted for the purpose of generating new knowledge. Must also not fall within any of the eight statutory exclusions in s 355–25(2).
Supporting R&D activitys 355–30, ITAA 1997
An activity directly related to a core R&D activity. Some categories — those that produce or relate to producing goods or services, or activities excluded from being core R&D — must additionally be conducted for the dominant purpose of supporting a core R&D activity. Routine maintenance, marketing, sales, and general administration do not qualify.
Dominant purposes 355–30(2), ITAA 1997
Your prevailing or most influential purpose. There can only be one. Required to qualify certain activities as supporting R&D activities. Assessed from project plans, board minutes, budget approvals, and time records — not from a retrospective characterisation.
Internal administration software exclusions 355–25(2)(h), ITAA 1997
Software developed for the dominant purpose of internal administration of the company's business is excluded from being a core R&D activity. The single biggest compliance risk for software companies — payroll, ERP, CRM, and invoicing systems built for internal use are caught. Document who the software is for and the dominant purpose of its use before claiming.
Apportionments 355–205, ITAA 1997
Where an employee, contractor, or asset is used for both R&D and non-R&D activities, you must apportion the expenditure on a reasonable, documented, and consistently applied basis. "We estimated 50% of developer time was R&D" without underlying time records is not acceptable — reviewers will ask what contemporaneous data the estimate was based on.
At-risk ruleTR 2021/5; s 355–405, ITAA 1997
Expenditure can only be notionally deducted as R&D if your company actually bears the financial risk. If the expenditure is funded, guaranteed, or reimbursed by another party — government grant, related party, or a contract that pays you regardless of outcome — it is not at risk and cannot be claimed.
Notional deductionSubdivision 355-B, ITAA 1997
The amount used to calculate your R&D tax offset. Includes eligible R&D expenditure, decline in value of assets used in R&D, and eligible CRC contributions. The offset is calculated by multiplying the notional deduction by the applicable offset rate. Not the same as an ordinary tax deduction.
Decline in values 355–210, ITAA 1997
Depreciation of tangible and intangible assets used in R&D activities. Apportioned where the asset has dual use. Must be supported by an asset register, evidence of R&D use (what asset, when, for which activity), and depreciation calculations.
Advance findingPart III, IR&D Act 1986
A legally binding decision by DISR that specified activities are, or would be, eligible core or supporting R&D activities. Provides certainty before you claim. Must be applied for — not automatic. DISR assesses your described activities against the legislative criteria.
Record retentions 262A, ITAA 1936
All R&D records must be retained for 5 years after the income year in which the expenditure is claimed. Applies to activity evidence, expenditure evidence, linkage evidence, and correspondence with DISR or the ATO. Store records in a system that preserves metadata — particularly creation dates and author information.
Competent professionalGuide to Interpretation, May 2024 p.16
A person who, in the relevant field, has knowledge, experience, and appropriate qualifications; keeps up to date with developments; and has access to worldwide knowledge and resources. The "unknown outcome" element is assessed from the perspective of such a person — not just from your team's perspective.
Disclaimer: This guide provides general information about R&DTI documentation and record-keeping to support self-assessment. It does not constitute formal tax advice, a legal opinion, or a registered tax agent service. The R&DTI is a self-assessment program — you remain responsible for ensuring you meet all legislative requirements before registering. Always refer to Division 355 of the Income Tax Assessment Act 1997, the Industry Research and Development Act 1986, s 262A of the ITAA 1936, and the Department of Industry, Science and Resources' Guide to Interpretation (May 2024) when assessing your records. Seek independent professional advice for your specific circumstances.