Industries

Proptech R&D is in the models and controls you test, not the cost of the building or the property group's back office

Proptech companies can claim the R&D Tax Incentive for experiments with an unknown outcome, such as valuation accuracy in thin markets or control of existing buildings' energy use. Constructing or fitting out a building, code and rating compliance, and software built mainly for a related property group's own administration do not qualify.

Reviewed by George WalchFacts last verified

Where the R&D usually sits

Proptech uncertainty comes in two kinds. Software and data questions: can a valuation model reach a stated error where comparable sales are thin; can room geometry and quantities be extracted from photographed or scanned plans to a measurable accuracy. Building science questions: can a control strategy cut energy in an existing building with partial sensor coverage without breaching comfort limits; how does a retrofit material behave in a real facade. Each has its own competent professional, a data or software engineer for the first and a building services engineer or building scientist for the second.

What usually qualifies

Proptech activities and how they usually fare
FactUsuallyWhy
Reaching a stated valuation error for regional properties with few comparable salesOften coreCore where published models cannot predict accuracy for that market
Extracting room geometry and quantities from photographed or scanned plans at a stated accuracyOften coreCore where accuracy on real, inconsistent drawings cannot be predicted from published methods
Model-based energy control for existing buildings with partial sensor coverageOften coreCore where building-specific behaviour makes savings unpredictable from published control results
Installing extra sensors and data pipelines in pilot buildings for those trialsSupportingDirectly enables the experiment; dominant purpose applies because the building keeps operating, and sensors are claimed through decline in value
Property management, leasing, inspection and CRM featuresNoUsually not core; routine software assembled from known patterns
Constructing or fitting out the building a trial runs inNoNot claimable as cost; expenditure to construct a building cannot be notionally deducted (s 355-225(1)(a)), even where an experiment runs inside it
Energy ratings, code compliance reports and certificationsNoExcluded from core as compliance (s 355-25(2)(f)); testing done mainly to feed a registered experiment can be supporting
Rent, fee or listing price experiments measured on yield or conversionNoExcluded from core as market testing (s 355-25(2)(a))

Illustrative example

An illustrative building control company's energy claim

A hypothetical company sells energy optimisation to office landlords. It hypothesised that a thermal model learned from 8 weeks of building management data would cut chiller energy by at least 15% in three existing towers of different ages, while holding zone temperatures between 21 and 24 degrees for 95% of occupied hours. The published control studies it reviewed used fully instrumented buildings, while these towers had partial sensor coverage. One tower failed the comfort limit because of a faulty damper that the data did not reveal. The pilots ran at the company's cost under agreements leaving the model and results with it. The control experiment is the core activity and the added sensors and data pipeline are supporting. Landlord dashboards and rating support are left out. This is an illustration, not a client.

The evidence you already have

Proptech teams hold model evaluation sets with error by region and property type, building management system exports, sensor logs with timestamps, trial agreements with building owners, and engineering notes on why a building behaved unexpectedly. The trial agreement matters more than teams often expect: it shows who controlled the pilot, who owns the model and data, and who paid, which is the conducted-for test.

Watch outs

Three risks are specific to property. Construction work is closely watched: TA 2017/2 targets whole-of-project claims, novel elements that in effect fulfil a building contract, and building expenditure. Proptech arms of property groups are exposed to the internal administration exclusion, which covers software used by connected and affiliate entities. And pilots in a client's building can leave the results with the client, so the R&D may not be conducted for you.

Is it eligible? training ml models, prototypes and tooling, materials testing

One company, two kinds of R&D

Many proptech companies are software businesses with a foot in the physical world. That gives them two distinct kinds of technical uncertainty, and they should not share a core activity.

The software side (valuation, plan analysis, search and matching) is judged against what an experienced data or software engineer could predict from published methods. AusIndustry's software guide treats building dashboards and integrating services by following vendor documentation as routine, which covers most property management and listings features.

The physical side (energy, comfort, materials, construction methods) is judged by a building services engineer or building scientist against engineering literature, codes and standards. Here the question is whether calculation and established practice could predict the result. If an engineer could model it confidently, there is no experiment.

The building can host the experiment, but its cost stays out

Construction claims are closely watched. Taxpayer Alert TA 2017/2 targets claims that treat a whole building project as R&D, claims where the "novel" element in effect just delivers the building contract, and claims that include building expenditure. A proptech pilot in a real building inherits the same scrutiny.

A building can still be the subject or the test site of a qualifying activity. What can qualify is a narrow question inside it: does this control strategy hold comfort while cutting energy, does this facade panel meet a thermal target in this climate. AusIndustry's built environment guidance notes that the "application of existing technology in a different context or location" can be core R&D, depending on the circumstances, so a known approach in a new building type is not automatically out.

Three separate rules then apply, and they answer different questions. Whether the activity is R&D is the ordinary core test. Whether a cost counts is a different matter: s 355-225(1)(a) stops expenditure to acquire or construct a building, or to extend, alter or improve one, being notionally deducted, even when an experiment runs inside it. And because the building keeps producing a service while the trial runs, work on it can only be supporting if its dominant purpose is the experiment.

Who the software is for

Section 355-25(2)(h) excludes software developed for the dominant purpose of internal administration by the company or a connected or affiliate entity. That matters for proptech businesses that sit inside property groups. A leasing or asset management platform built mainly to run the parent's own portfolio is internal administration for the group, even if it lives in a separate company. The same platform sold to unrelated agencies and landlords is a product. See internal administration software.

Group structures raise a second question. Where the parent funds the work and keeps the results, the R&D may be conducted for the parent. Taxpayer Alert TA 2023/4 describes a related pattern, where a group's R&D entity claims for work that an associate in substance funds, controls and exploits.

How proptech registrations usually split

  • Separate core activities for software questions and building science questions.
  • Supporting activities for pilot instrumentation, data labelling and test rigs, with any building that keeps operating apportioned.
  • Left out: construction and fit-out, compliance and ratings, property operations software, and pricing experiments.

Keep the pilot agreement with each building owner with the trial records. It is the document that shows your company controlled the work, paid for it and kept the results.

For physical prototypes and test panels, see prototypes and tooling and materials testing.

Frequently asked questions

Is our automated valuation model R&D?
Only the part where the result was genuinely unknown. Training a standard gradient boosted model on listing and sales data is applying a known method. Reaching a stated error on regional properties with few comparable sales, where published approaches do not, and testing why, can be a core activity.
We are a developer trialling a new construction method on one of our projects. What can we claim?
The specific experimental question, for example whether a connection detail achieves a load or acoustic target that engineering calculation cannot predict. Not the project. TA 2017/2 warns against claiming a building as one activity, and the cost of the building itself is excluded.
Do energy ratings or code compliance reports count?
Not as core. Obtaining a rating or showing a design meets the code applies a published protocol to show compliance, which s 355-25(2)(f) excludes from core. Testing done mainly to feed a registered experiment can be supporting work.

Sources

  1. AusIndustry: Built environment sector guide
  2. business.gov.au: Excluded R&D activities under the R&DTI (internal administration software)
  3. AusIndustry: Software development sector guide

The R&D Tax Incentive is a self-assessment program. This page is general information, not tax, legal or financial advice; eligibility depends on your circumstances and you should seek independent advice for them.