REFERENCE
R&D Tax Sector Benchmarks
Sector-by-sector benchmarking for the Australian R&D Tax Incentive. Direct (core) and supporting/indirect R&D ranges as a percentage of total business expenses, drawn from ATO RDTI program data, AusIndustry sector guidance and ABS innovation statistics. These bands are educational - sitting outside the range does not by itself imply ineligibility.
Green
Within commonly observed direct-R&D range.
Amber
Outside common range but still potentially valid.
Red
Materially outside typical pattern - may benefit from extra substantiation.
Supporting
Indirect activities undertaken for the dominant purpose of supporting a core activity.
Healthcare & Medical
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
Dental clinics Most clinical dentistry is routine treatment and excluded. Bona-fide R&D usually arises only where the practice runs structured experimental protocols (new material trials, digital workflow validation, custom implant design). | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Dental practices typically allocate 0–3% of expenses to direct R&D. Anything materially higher is not necessarily wrong, but it usually reflects formal experimental protocols, IP development or a distinct device/software arm rather than day-to-day clinical work. | Section 355-25(2)(b) excludes activities whose outcome can be known from current knowledge. Routine restorations, scans and treatments fall outside core R&D. |
General medical practice GP clinics rarely generate experimental work other than registered clinical trials or distinct software/process R&D arms. | 0%50%100% | 0–3% | ≤7% | >7% | 0–2% | Day-to-day general practice is consultative care, not R&D. Higher allocations typically only appear where the clinic is running formal trials or developing a separate digital health / triage product. | ATO TR 2021/5 emphasises hypothesis-led experimentation. Diagnosis-and-treatment workflows are not core R&D activities. |
Specialist clinics Surgical and specialist clinics occasionally run formal protocols, device pilots or novel procedure development. | 0%50%100% | 0–5% | ≤12% | >12% | 0–3% | Specialist practices can have legitimately higher allocations than GPs where they run protocol-driven experimentation, but most of the practice's expense base is still routine clinical care. | Eligibility hinges on whether the procedure is genuinely novel and hypothesis-led, not merely complex. |
Radiology / imaging AI-assisted reading, novel imaging protocols and reconstruction algorithms can drive a small but real R&D layer. | 0%50%100% | 0–6% | ≤15% | >15% | 1–4% | Routine scanning is not R&D. Higher allocations are plausible where the practice is developing AI tooling or validating new acquisition protocols. | Hardware purchases are not R&D; only the experimental development around them may qualify. |
Pathology Assay development, novel reagent validation and LIMS algorithm work can be eligible; routine testing throughput is not. | 0%50%100% | 1–8% | ≤20% | >20% | 1–5% | Production testing volume is excluded. Higher percentages usually reflect a discrete assay-development or molecular-diagnostics arm. | Routine production once the underlying knowledge is established falls outside the s355-25 core activity definition (no remaining technical unknown). |
Allied health (physio, OT, podiatry) Allied health is overwhelmingly service delivery. Eligible R&D usually means a digital tool, novel exercise protocol with formal measurement, or a custom device. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Patient consultations and standard rehabilitation protocols are commercial-as-usual. Material R&D allocations generally indicate a product or platform built alongside the clinic. | Customary clinical practice is excluded; documented experiments with measurable outcomes may qualify. |
MedTech (device development) Pre-TGA device companies are R&D-dominant. Post-approval intensity falls quickly as manufacturing and commercial functions scale. | 0%50%100% | 30–70% | ≤85% | >85% | 5–15% | Device companies pre-market routinely sit at 50–80% direct R&D. Anything materially lower may indicate under-claim of supporting engineering and regulatory work. | Verification & validation testing required for regulatory approval is generally a supporting activity, not core R&D. |
Biotechnology Clinical-stage biotech runs near-pure R&D until commercial revenues appear. Manufacturing scale-up shifts the ratio sharply. | 0%50%100% | 40–85% | ≤92% | >92% | 3–10% | Pre-commercial biotech allocations above 80% are entirely typical. The relevant compliance focus is contemporaneous documentation of each experiment, not the headline percentage. | Clinical trial activity is supporting where it validates eligible core experimentation rather than routine safety/efficacy. |
Pharmaceuticals (commercial-stage) Once products are on-market, sales, distribution and manufacturing dilute R&D intensity. Reformulation and next-generation pipeline drives what remains. | 0%50%100% | 10–35% | ≤55% | >55% | 3–10% | Mature pharma businesses commonly land in the teens. Higher percentages usually indicate either an active pipeline cohort or a pre-commercial subsidiary structure. | Bioequivalence and routine reformulation often fall outside eligible R&D unless genuine technical uncertainty is documented. |
Medical devices — manufacturing Established device manufacturers split between production (excluded) and next-gen development (eligible). | 0%50%100% | 5–20% | ≤40% | >40% | 2–8% | Production volume dilutes intensity. Higher percentages are plausible where a discrete next-generation device program is underway. | Routine production line activities fall outside the s355-25 core activity definition. |
Digital health / health software Behaves like SaaS but with additional regulatory and clinical-validation supporting activity. | 0%50%100% | 25–70% | ≤85% | >85% | 3–10% | Pre-revenue digital health platforms regularly exceed 60% direct R&D. Once recurring revenue scales, intensity drops in line with mature SaaS. | Validation of algorithm performance against clinical endpoints can qualify if hypothesis-led. |
Aged care innovation Care delivery is excluded. Eligible R&D usually sits inside a tech subsidiary or formal pilot programme. | 0%50%100% | 0–4% | ≤10% | >10% | 0–3% | Standard care delivery is not R&D. Higher allocations typically reflect a software arm or government-co-funded pilot, not the operating facility itself. | Care-services revenue and staffing should be excluded from notional R&D unless directly tied to an experimental protocol. |
Technology
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
SaaS (pre-revenue / early-stage) Engineering headcount dominates the cost base; revenue is yet to scale operations and sales. | 0%50%100% | 50–80% | ≤90% | >90% | 5–15% | Pre-revenue SaaS commonly sits between 60–85% direct R&D. Higher allocations are plausible for pure R&D vehicles, but full-company allocations above 95% usually need apportionment review. | AusIndustry guidance recognises iterative software development where hypothesis-driven and not merely customisation. |
SaaS (revenue-generating) Sales, marketing, support and infrastructure operations dilute intensity once ARR scales. | 0%50%100% | 15–40% | ≤60% | >60% | 3–12% | Revenue-stage SaaS commonly settles between 20–40% direct R&D. Higher allocations are plausible during major platform re-architectures. | Routine maintenance, bug fixes, customer-specific config and BAU DevOps are excluded. |
Custom software development services Client-funded build work is typically excluded; only internal IP or genuinely novel methods qualify. | 0%50%100% | 3–15% | ≤30% | >30% | 1–5% | Service-revenue dev shops commonly land in the single digits. Material allocations usually point to a separately-funded internal product the agency owns. | Where the client owns the IP and bears the risk, the work is generally not on own behalf and not eligible. |
Cybersecurity Detection-engine and protocol R&D qualifies; SOC operations and managed-service delivery do not. | 0%50%100% | 20–55% | ≤75% | >75% | 3–10% | Product-oriented cyber firms regularly sit 30–55%. Higher percentages need a clear distinction between platform R&D and security-operations service delivery. | Threat-intel collection and incident response are operational, not experimental. |
AI / ML Model architecture, training-regime experimentation and evaluation methodology drive direct R&D. | 0%50%100% | 35–75% | ≤88% | >88% | 5–15% | AI-first companies routinely run above 50% direct R&D. Data-labelling and inference operations should be classified as supporting, not core. | Applying off-the-shelf models without genuine algorithmic uncertainty is typically not eligible. |
Data platforms / analytics Pipeline and query-engine R&D qualifies; dashboard configuration and data-ops do not. | 0%50%100% | 20–50% | ≤70% | >70% | 3–10% | Platform-stage data businesses commonly sit 25–50%. Reporting, BI and ETL maintenance are excluded. | Performance optimisation can qualify only where measurable hypotheses and uncertain outcomes are documented. |
FinTech Core ledger, risk-engine and protocol work qualifies; KYC, compliance ops and licensing do not. | 0%50%100% | 25–55% | ≤75% | >75% | 5–15% | Engineering-heavy fintechs sit 30–55%. Compliance, customer support and account-management costs should not appear in core R&D. | Regulatory compliance work is excluded; underlying engineering uncertainty may still qualify. |
PropTech Platform engineering qualifies; brokerage operations and listings do not. | 0%50%100% | 20–55% | ≤75% | >75% | 3–12% | Product-led PropTech sits in line with SaaS. Marketplace operations should be carved out of R&D. | Inventory acquisition and transaction execution are commercial-as-usual. |
HealthTech (non-clinical) Behaves like SaaS plus regulatory validation supporting activity. | 0%50%100% | 25–65% | ≤82% | >82% | 5–15% | Pre-commercial HealthTech often exceeds 60%. Post-launch operations dilute intensity in line with mature SaaS. | Clinical validation can be supporting; routine usability testing is not eligible. |
Deep tech / R&D-heavy startups Companies in quantum, photonics, advanced materials, novel semiconductors etc. often have minimal non-R&D spend. | 0%50%100% | 50–90% | ≤95% | >95% | 3–12% | Deep-tech businesses regularly sit at 80%+ direct R&D and that is entirely consistent with their commercial model. The compliance focus is on per-experiment documentation, not the headline ratio. | Capital equipment purchases are not R&D; their consumable use during experiments may be. |
Hardware Prototype iteration, EVT/DVT cycles and test-rig builds drive intensity; production tooling is excluded. | 0%50%100% | 20–55% | ≤75% | >75% | 5–15% | Pre-mass-production hardware sits 30–55%. Lower allocations after launch are expected. | Tooling and production-run costs fall outside the s355-25 core activity definition; reverse-engineering existing products is separately excluded under s355-25(2)(g). |
Embedded systems / IoT Firmware, power-management and connectivity stack R&D dominate pre-launch. | 0%50%100% | 20–55% | ≤75% | >75% | 5–15% | Embedded product companies sit similarly to hardware; integration with existing modules without uncertainty is not core R&D. | Use of off-the-shelf SDKs without modification rarely qualifies. |
Cloud infrastructure Custom orchestration, scheduling and networking R&D qualifies; capacity operations and customer support do not. | 0%50%100% | 20–55% | ≤75% | >75% | 5–15% | Platform companies sit in line with SaaS. Compute resale margin and account ops are not R&D. | Provisioning infrastructure for paying customers is operational, not experimental. |
Enterprise software Field engineering, customer-specific config and integration work is generally excluded. | 0%50%100% | 15–40% | ≤60% | >60% | 3–12% | Mature enterprise software sits 15–40%. Customer-specific deliverables should be separated from core platform R&D. | Customisation funded by a customer is generally not on own behalf. |
Pre-revenue tech startup (general) Headcount is engineering-heavy; commercial functions are minimal. | 0%50%100% | 45–85% | ≤92% | >92% | 5–15% | Pre-revenue tech startups frequently sit above 70% direct R&D — that is consistent with the business model. Documentation of hypotheses is what matters at audit, not the percentage. | Founder time can be claimed at reasonable apportionment if directly engaged in experimental activity. |
Product development firms / studios Most output is client-funded; only the studio's own IP, frameworks and reusable platforms may qualify. | 0%50%100% | 5–20% | ≤40% | >40% | 2–8% | Service revenue is excluded. Material allocations usually require a clearly delineated internal-product or IP-development stream. | Confirm 'on own behalf' for each project — IP ownership and financial risk are the key tests. |
Engineering & Manufacturing
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
Advanced manufacturing Process and product development qualifies; serial production is excluded. | 0%50%100% | 8–25% | ≤45% | >45% | 3–10% | Advanced manufacturers commonly sit 10–25%. Higher percentages usually reflect a dedicated NPD or process-engineering team. | Routine production once knowledge is established is not a core activity under s355-25 (the technical unknown has been resolved). |
Industrial manufacturing Capital-intensive operations with concentrated, episodic R&D around new product lines or major retools. | 0%50%100% | 2–10% | ≤20% | >20% | 1–5% | Steady-state industrial manufacturing sits in the low single digits. Higher percentages are plausible during a specific line-development or new-process trial year. | Capital equipment is depreciable; only the experimental use of it is potentially eligible. |
Food manufacturing Recipe development, shelf-life trials and process scale-up may qualify; existing-product production runs do not. | 0%50%100% | 2–10% | ≤20% | >20% | 1–5% | Food manufacturers commonly sit 3–10%. Reformulation projects need documented hypotheses and outcomes, not just R&D-coded GL accounts. | Routine quality control, regulatory labelling and packaging changes are excluded. |
Beverage manufacturing New-product trials and fermentation/process experimentation drive direct R&D. | 0%50%100% | 2–10% | ≤20% | >20% | 1–5% | Routine production runs and seasonal SKU rotations are not R&D. | Sensory panel evaluation may be supporting where it validates eligible experimentation. |
Chemical manufacturing Formulation, catalyst and process R&D can sustain a meaningful intensity layer. | 0%50%100% | 4–15% | ≤30% | >30% | 2–7% | Chemical manufacturers commonly sit 4–15%. Pilot-plant operations may be supporting where they validate eligible core experiments. | Scale-up trials must be hypothesis-led to remain eligible; routine bulk production is not. |
Packaging Sustainable-materials R&D and barrier-performance trials drive eligible activity. | 0%50%100% | 2–10% | ≤20% | >20% | 1–5% | Design refresh and printing changes are excluded. Material-substitution programs with measurable performance hypotheses may qualify. | Cosmetic redesign is excluded; barrier-property experimentation can qualify. |
Industrial / mechanical engineering Project-based, with R&D concentrated in prototyping and validation phases. | 0%50%100% | 3–12% | ≤25% | >25% | 1–6% | Routine detailed engineering is excluded. Eligible activity centres on genuinely novel mechanisms or first-of-kind builds. | Standard engineering design from established principles is not eligible (AusIndustry sector guidance). |
Electrical engineering Embedded firmware and novel power-electronics work may qualify; installation and commissioning do not. | 0%50%100% | 3–12% | ≤25% | >25% | 1–6% | Field installation work is excluded. R&D usually sits with a small product or controls-engineering team. | Routine commissioning and switchboard wiring are excluded. |
Mining equipment / METS Equipment design, autonomy software and processing innovation drive intensity. | 0%50%100% | 5–20% | ≤40% | >40% | 2–8% | Field-service revenue is excluded. R&D usually sits within product development arms. | On-site trials may be supporting; routine maintenance is not eligible. |
Materials science Pre-commercial materials companies are R&D-dominant; scale-up shifts the ratio. | 0%50%100% | 25–70% | ≤85% | >85% | 3–12% | Pre-revenue materials companies regularly exceed 60%. Production from validated formulations is excluded. | Characterisation testing may be supporting where it validates hypothesis-led experimentation. |
Prototyping / fabrication firms Most output is client-funded; only own-IP or methodology R&D qualifies. | 0%50%100% | 3–15% | ≤35% | >35% | 2–7% | Build-to-print work is excluded. Material allocations usually require a clear internal-IP stream. | Confirm 'on own behalf' on a project basis. |
Industrial product design Pure styling/aesthetic work is excluded; performance-driven design with measurable hypotheses may qualify. | 0%50%100% | 3–15% | ≤30% | >30% | 1–5% | Industrial design fees billed to clients are not R&D. Eligible work usually involves an internal product line or genuinely novel engineering. | Marketability work falls under the market-research exclusion in s355-25(2)(a); purely aesthetic work generally fails the new-knowledge limb of s355-25. |
Construction & Property
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
Residential construction Residential build is overwhelmingly routine. Genuine R&D usually involves novel materials or modular systems. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Standard residential construction is not R&D. Higher allocations usually reflect a discrete modular, prefab or sustainability R&D programme. | AusIndustry sector guidance for Building & Construction is explicit that customary trade practice is excluded. |
Commercial construction Project-based, with rare experimental work around novel facade, structural or services systems. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Routine commercial build work is excluded. Eligible R&D arises only where there is genuine technical uncertainty not resolvable with established engineering. | First-of-type problems within Australia can qualify if hypothesis-led and documented. |
Modular / prefab construction Connection systems, load-path verification and factory-process R&D drive eligible activity. | 0%50%100% | 3–12% | ≤25% | >25% | 1–6% | Modular firms commonly sit higher than traditional builders. Repeat production of a validated module is excluded. | Recurrent production runs of a validated module are excluded. |
Construction technology (ConTech) Software, robotics and digital-twin platforms behave like SaaS / hardware. | 0%50%100% | 25–65% | ≤82% | >82% | 5–15% | Product-led ConTech sits in line with SaaS or hardware ranges, not with the construction operating business. | Pure construction services delivered alongside the platform must be carved out. |
Engineering consultancies (civil/structural) Client-funded design work is excluded; internal methodology or tooling R&D may qualify. | 0%50%100% | 1–6% | ≤15% | >15% | 0–3% | Fee-for-service design output is not R&D. Material allocations point to internal tooling, BIM frameworks or proprietary modelling work. | Where the client owns IP and bears risk, the work is generally not on own behalf. |
Architecture Practice work is excluded; eligible activity is rare and usually research-oriented. | 0%50%100% | 0–4% | ≤10% | >10% | 0–2% | Design service revenue is not R&D. Eligible work is typically a separately-funded research collaboration. | Marketability work is excluded by s355-25(2)(a) (market research/sales promotion); purely aesthetic work generally fails the new-knowledge limb. |
Property development Land acquisition, sales and marketing dominate the cost base. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Property development costs are predominantly capital and commercial. R&D allocations should be modest and project-specific. | Holding costs and finance charges are not R&D. |
Sustainability / green building innovation Novel material trials and energy-performance experiments drive eligible activity. | 0%50%100% | 5–25% | ≤45% | >45% | 2–8% | Compliance with NCC/BCA targets alone is not R&D. Experimentation beyond established performance is required. | Compliance work is excluded; experimentation beyond established performance can qualify. |
Agriculture & Primary
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
AgTech Software, sensors and robotics R&D behave like SaaS / hardware. | 0%50%100% | 25–65% | ≤82% | >82% | 3–12% | Product-led AgTech sits in line with SaaS or hardware. On-farm services should be carved out. | Field trials are supporting where they validate eligible core experiments. |
Broadacre farming Production farming is excluded. Eligible R&D is usually a formal trial co-funded by an RDC. | 0%50%100% | 0–5% | ≤12% | >12% | 0–3% | Routine cropping operations are not R&D. Higher allocations usually accompany a documented multi-season trial. | RDC / Grains Research co-funded trials may qualify where hypothesis-led. |
Livestock Standard husbandry is excluded; genetics, nutrition trials or methane-reduction work may qualify. | 0%50%100% | 0–5% | ≤12% | >12% | 0–3% | Production farming is not R&D. Material allocations need a documented trial protocol. | Selective breeding without controlled experimental design is not eligible. |
Aquaculture Disease management, feed formulation and recirculating-system design drive eligible activity. | 0%50%100% | 3–15% | ≤30% | >30% | 1–6% | Standard grow-out operations are excluded. Material allocations need formal trial protocols. | Stock losses during experimental trials may be eligible expenditure if directly tied to the experiment. |
Viticulture / wine New-variety trials, fermentation experimentation and disease-resistance trials drive eligibility. | 0%50%100% | 1–8% | ≤18% | >18% | 0–4% | Established varietal production is not R&D. Genuine trials need documented hypotheses and outcomes. | Vintage-to-vintage variation is not R&D unless tied to a controlled experimental design. |
Crop science Trait-development companies are R&D-dominant. | 0%50%100% | 30–70% | ≤85% | >85% | 3–10% | Pre-commercial crop-science businesses regularly exceed 60% direct R&D. | Field trial expenditure is supporting where it validates hypothesis-led experimentation. |
Food innovation / NPD Alternative-protein, fermentation and novel-format work drive eligible activity. | 0%50%100% | 5–25% | ≤45% | >45% | 2–8% | Routine recipe variation is excluded; first-of-kind formats and ingredient platforms may qualify. | Regulatory submissions (FSANZ) are not themselves R&D. |
Irrigation technology Controls, sensors and water-management R&D behave like hardware / IoT. | 0%50%100% | 10–35% | ≤55% | >55% | 2–8% | Installation services are excluded; product R&D dominates eligible activity. | Field deployment is supporting where it validates eligible core experimentation. |
Mining & Resources
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
Mining operations Mining production is largely excluded by carve-outs; only narrowly-defined experimental processing work may qualify. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Standard mining production is not R&D. Higher allocations usually point to a specific processing or recovery experiment. | Prospecting, exploring and drilling for minerals or petroleum for the purpose of locating or quantifying deposits is excluded under s355-25(2)(b). |
Mineral exploration Novel geophysical techniques and ore-body characterisation methods may qualify. | 0%50%100% | 2–10% | ≤25% | >25% | 0–4% | Routine drilling for resource estimation is not R&D. New technique trials may qualify if hypothesis-led. | Prospecting and routine drill-out are excluded; novel technique trials may qualify. |
Energy generation (utility-scale) Operating utilities have low intensity; R&D is concentrated in pilot projects. | 0%50%100% | 0–5% | ≤15% | >15% | 0–3% | Day-to-day generation operations are not R&D. Pilot deployments need clear experimental scope. | Standard commissioning is excluded; novel first-of-kind systems may qualify. |
Oil & gas Subsea, EOR and process-engineering R&D may qualify; production is excluded. | 0%50%100% | 0–5% | ≤12% | >12% | 0–3% | Production is not R&D. Eligible work is usually a discrete technology-pilot project. | Activities related to the extraction of natural resources are explicitly excluded. |
Renewable energy / storage Cell, inverter, storage chemistry and grid-integration R&D drives intensity. | 0%50%100% | 15–50% | ≤75% | >75% | 3–12% | Pilot-stage renewables commonly sit 40–60%. Commissioning and operations dilute the ratio sharply once commercial. | EPC delivery work is excluded; technology development around it may qualify. |
Environmental technology Remediation methods, monitoring systems and circular-economy process R&D drive eligibility. | 0%50%100% | 10–40% | ≤60% | >60% | 3–10% | Standard environmental consulting is excluded; method-development work may qualify. | Compliance reporting is not R&D. |
Professional Services
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
Accounting firms Fee-for-service work is excluded; only internal product or methodology R&D may qualify. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Service delivery is not R&D. Material allocations almost always indicate an internal software product the firm owns. | Confirm IP ownership and 'on own behalf' for internal tooling claims. |
Legal firms Practice work is excluded; LegalTech subsidiaries can have separate ranges. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Legal services are not R&D. Higher percentages usually indicate a distinct LegalTech entity. | Knowledge management and precedent libraries are not R&D. |
Management consulting Most consulting engagements are excluded; proprietary platforms or methodologies may qualify. | 0%50%100% | 0–4% | ≤10% | >10% | 0–2% | Client-delivered consulting is not R&D. Eligible work usually sits within a productised IP stream. | Knowledge synthesis from public sources is not R&D. |
Marketing agencies Marketing campaign delivery is excluded; MarTech tooling development may qualify. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Creative production and campaign execution are not R&D. Eligible work is usually internal platform development. | Market research, market testing and sales promotion are excluded under s355-25(2)(a). |
Creative agencies Creative output is excluded; toolchain and pipeline R&D may qualify. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Creative deliverables are not R&D. Material allocations need a clearly defined internal-tools program. | Purely aesthetic work generally fails the new-knowledge limb of s355-25; toolchain R&D may still qualify on its own merits. |
HR services / HRTech Pure HR services are low intensity; HRTech product arms behave like SaaS. | 0%50%100% | 5–30% | ≤55% | >55% | 2–10% | If you operate primarily as a services business, expect single-digit allocations; HRTech products follow SaaS ranges. | Recruitment and placement services are not R&D. |
General business services Most BPO / outsourced services work is excluded. | 0%50%100% | 0–3% | ≤8% | >8% | 0–2% | Service delivery is not R&D. Material allocations need a distinct internal product or platform. | Internal process improvement without technical uncertainty is not eligible. |
Retail, E-commerce & Consumer
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
E-commerce (pure-play) Inventory, marketing and fulfilment dominate; only platform and logistics-tech R&D may qualify. | 0%50%100% | 1–8% | ≤18% | >18% | 0–4% | Merchandising, advertising and fulfilment are excluded. Higher percentages usually indicate a meaningful internal-engineering team building proprietary platforms. | Off-the-shelf platform configuration (Shopify, BigCommerce) is not R&D. |
Consumer products (DTC brand) Brand and marketing dominate; product-formulation R&D may qualify. | 0%50%100% | 1–8% | ≤18% | >18% | 0–4% | Brand activity is not R&D. Eligible allocations usually sit with formulation or material-development work. | Aesthetic redesign is excluded; performance experimentation may qualify. |
FMCG Mass-production COGS dilutes intensity. Discrete NPD pipelines may qualify. | 0%50%100% | 1–7% | ≤15% | >15% | 0–4% | Sales and distribution are excluded. R&D usually centres on a small NPD team and pilot lines. | Routine line-extensions and pack-format changes are not eligible. |
Retail technology POS, inventory and supply-chain platform R&D behave like SaaS. | 0%50%100% | 20–55% | ≤75% | >75% | 3–12% | Product-led retail tech sits in line with SaaS. Reseller and implementation revenue must be carved out. | Customer-specific configuration is generally excluded. |
Warehousing / 3PL innovation Robotics and WMS development may qualify; pick-and-pack operations do not. | 0%50%100% | 2–10% | ≤25% | >25% | 1–5% | Operations are excluded. Material allocations usually accompany a robotics or software development arm. | Standard 3PL operations are commercial-as-usual. |
Logistics innovation Routing, telematics and platform R&D may qualify; freight operations do not. | 0%50%100% | 2–12% | ≤28% | >28% | 1–6% | Carrier operations are excluded. Eligible work usually sits in a software or hardware development team. | Network operations are not R&D. |
Education
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
EdTech Behaves like SaaS, often with additional pedagogy / efficacy validation. | 0%50%100% | 25–65% | ≤82% | >82% | 5–15% | Product-led EdTech sits in line with SaaS. Content production should be carved out from core R&D. | Content creation is generally not R&D; algorithmic personalisation may qualify. |
Training innovation Course delivery is excluded; novel methodology research or platform R&D may qualify. | 0%50%100% | 0–4% | ≤10% | >10% | 0–3% | Training delivery is not R&D. Material allocations need a distinct technology or measurement programme. | Curriculum development is not eligible under s355-25(2). |
Learning management systems Platform engineering behaves like SaaS. | 0%50%100% | 20–55% | ≤75% | >75% | 3–12% | Implementation services and content licensing should be carved out from core R&D. | Customer-specific configuration is excluded. |
Transport & Logistics
| Sub-sector | Direct R&D range | Typical | Elevated | Unusual | Supporting | Educational disclaimer | ATO / AusIndustry context |
|---|---|---|---|---|---|---|---|
Fleet / mobility technology Vehicle telematics, autonomy stack and routing R&D drive intensity. | 0%50%100% | 20–55% | ≤75% | >75% | 3–12% | Fleet operations are excluded. Product engineering may sit in line with hardware / SaaS. | Driver hours and fleet maintenance are not R&D. |
Logistics software Routing-engine and optimisation R&D qualifies; implementation services do not. | 0%50%100% | 20–55% | ≤75% | >75% | 3–12% | Implementation and managed services should be carved out. | Customer-specific configuration is generally excluded. |
Supply chain technology Planning, forecasting and visibility platform R&D behave like SaaS. | 0%50%100% | 15–50% | ≤70% | >70% | 3–12% | Consulting and integration revenue should be carved out from core R&D. | Off-the-shelf forecasting package configuration is not R&D. |
1. Cross-industry principles
- Eligibility is activity-based, not industry-based
Every claim is tested against the s355-25 core activity definition and s355-30 supporting activity definition. Sector benchmarks are a sense-check, not an eligibility test.
- High R&D % is normal in some industries
Pre-revenue SaaS, biotech, deep tech and materials science routinely run above 60% direct R&D and that is consistent with their business model.
- Low R&D % can still be valid
Service businesses, construction firms and clinical practices may genuinely have only a small experimental layer. That layer can still be eligible when properly documented.
- Direct vs supporting must be separated
Core activities are the experiments. Supporting activities are everything undertaken for the dominant purpose of supporting a core activity. They are claimed differently and must be classified separately.
- Contemporaneous documentation matters more than headline %
AusIndustry reviews focus on whether each activity was hypothesis-led, involved technical uncertainty, and was conducted in a scientific manner — not on the company-level ratio.
- Exclusions are specific and material
Market research, sales promotion, routine data collection, social science research, prospecting, mining production and bid preparation are all explicitly excluded under s355-25(2).
2. Software flagging rules
| Rule | Behaviour |
|---|---|
| intensity within [typicalLow, typicalHigh] | GREEN — confirm and move on. |
| intensity below typicalLow | AMBER — surface under-claim prompt (review supporting activities / apportionment). |
| intensity (typicalHigh, elevatedHigh] | AMBER — surface 'above typical but plausible' message with documentation prompt. |
| intensity > elevatedHigh | RED — consultative banner, suggest specialist review, never block submission. |
| sector = service business and intensity > 15% | Prompt the user to confirm IP ownership and 'on own behalf' test. |
| supporting % > directTypicalHigh | Prompt: supporting activity allocations rarely exceed direct activity — verify classification. |
3. Disclaimer template library
greenWithinBand
Based on commonly observed patterns in Australian businesses operating in this sector, your allocation sits within the typical band. ATO and AusIndustry guidance focuses on the nature of eligible activities rather than industry averages alone — keep contemporaneous documentation for each experiment.
amberBelow
Your allocation is below the commonly observed range. This is not an issue — it can simply indicate under-claim of eligible supporting activities or apportioned overheads. Review whether time apportionment, on-costs and indirect activities have been captured.
amberAbove
Your allocation is above commonly observed ranges. This does not necessarily indicate an issue. Some businesses legitimately undertake significantly more R&D than peers, particularly where substantial experimentation, proprietary development or technical uncertainty exists. Additional context or documentation may help explain the commercial and technical drivers of the activity.
redMaterial
Your allocation is materially outside the typical pattern for this sector. There may be perfectly valid commercial reasons for this allocation. Specialist review or additional substantiation may help support the position; this does not by itself imply ineligibility.
serviceBusinessReminder
Where the work is funded by a client who owns the resulting IP and bears the technical risk, the activity is generally not 'on own behalf' and is therefore not eligible.
productionExcluded
Routine production once the underlying knowledge is established is not a core R&D activity under s355-25 — the technical unknown has been resolved. Only the experimental, hypothesis-led portion of activity should be classified as core R&D.
directVsSupporting
Direct (core) R&D activities are the experiments themselves. Supporting activities are those undertaken for the dominant purpose of supporting a core activity — they should be classified separately and may have their own apportionment.
4. Edge case guidance
- Pure R&D vehicle (no commercial revenue)
Intensity will sit near 100%. Apportionment is still required to ensure non-R&D costs (e.g. statutory accounting) are excluded.
- Group structure with shared services
Compute intensity at the R&D-conducting entity level, not the group. Recharge agreements must be at arm's length.
- Government grant co-funded project
Net the eligible expenditure against any feedstock / clawback amounts. Grant accounting can distort intensity for one period only.
- Overseas R&D activity
Overseas activity requires an Advance Overseas Finding. Excluded from the simple intensity calc unless covered by an AOF.
- Acquired in-process R&D
Capitalised IPR&D is not eligible expenditure — exclude from notional R&D when computing intensity.
- Pre-incorporation expenses
Pre-incorporation costs are generally not eligible. Exclude when computing intensity.
5. Startup vs mature company differences
- Pre-revenue startups routinely sit 30–60 percentage points above the same sector's mature businesses.
- Apply the early-stage sub-sector benchmark in the first 2–3 years; switch to the revenue-generating benchmark once recurring revenue exceeds direct R&D spend.
- Founder time can be apportioned reasonably if directly engaged in experimental activity, and is usually a meaningful chunk of pre-revenue R&D claims.
6. Common false positives
- Service-revenue businesses with a single internal product look like high-intensity firms in the GL — separate the cost base before computing intensity.
- Founder salary apportionment can lift intensity to above 90% in pre-revenue years — that is normal, not a red flag.
- Grant-funded pilot projects may push a single-year intensity high; benchmark across the program period rather than one year.
- Consolidated groups: include only the operating cost base of the entities actually conducting the R&D when computing intensity.
7. Industries where high R&D % is normal
- Biotechnology (pre-commercial)
- Deep tech / materials science / quantum
- MedTech pre-TGA
- Pre-revenue SaaS and AI-first startups
- Crop science / breeding programs
8. Industries where lower R&D % may still be valid
- Dental, GP and allied health practices running structured protocols
- Construction businesses with first-of-kind technical challenges
- Food manufacturers with discrete reformulation programs
- Engineering consultancies investing in internal IP / tooling
- FMCG businesses with concentrated NPD pipelines
9. UI tooltip & message library
intensityField
Direct R&D intensity = notional R&D deductions ÷ total company expenses. We use it to benchmark, not to determine eligibility.
green
Within the range commonly observed for this sub-sector. Keep contemporaneous documentation for each experiment.
amberBelow
Below the typical range. Not a problem — review whether supporting activities and apportioned overheads have been captured.
amberAbove
Above the typical range. Still plausible depending on your business model. Additional documentation may help support the position.
red
Materially outside the typical pattern for this sub-sector. Not a disqualifier — specialist review or additional substantiation may help.
directVsSupporting
Direct = the experiment itself. Supporting = activities whose dominant purpose is supporting a core activity (e.g. regulatory validation, supporting engineering).