Quick answer: To write an R&D technical narrative that survives an ATO review, structure it as a systematic experiment a reviewer can follow: state the technical unknown that couldn't be known in advance, the hypothesis you tested, the systematic experiment(s) you ran, your observation and evaluation of results, and the logical conclusion. Describe a test, not a build.
25 June 2026 — the 2026–27 Federal Budget announced R&DTI changes scheduled to start 1 July 2028; this article describes the current rules unless stated otherwise.
Almost every R&D Tax Incentive article warns you about the same thing: claims fall over at review because the documentation describes a software project, not an experiment. The cautionary cases get cited endlessly. What you rarely see is the part that actually helps — how to write the description in the first place.
That gap matters because the technical narrative is the single document an AusIndustry or ATO reviewer reads first. If it reads like a product update — "we built a new platform, it was hard, we shipped it" — there is nothing in it to assess against the law. If it reads like a structured experiment, the reviewer can follow your reasoning and check it against the eligibility tests.
This article gives you the structure of a defensible core activity description, a short before/after example, and a note on supporting activities. It is general information from a Registered Research Service Provider, not tax advice — eligibility depends on your circumstances and you should self-assess and seek your own advice.
What the law is actually asking for
Under the R&D Tax Incentive, a core R&D activity is an experimental activity whose outcome could not be known or determined in advance on the basis of current knowledge, and which is conducted for the purpose of generating new knowledge. The law requires that outcome to be worked out through a systematic progression of work — from hypothesis, to experiment, to observation and evaluation, to logical conclusions (Income Tax Assessment Act 1997, s 355‑25).
AusIndustry frames the same idea in plainer language: you must be testing a hypothesis using a systematic progression based on principles of established science (business.gov.au, Check if you are eligible for the R&DTI). Your narrative succeeds when a reader can see all five of those elements without having to guess. It fails when one or more is missing — most often the unknown and the experiment.
The five-part template
Write your core activity description so a reviewer can find each of these, ideally under its own heading. Copy the table below and fill the right-hand column for each core activity:
A technical narrative should describe the experiment, not the whole product build.
Part of the narrativeWhat to writeTechnical unknownThe specific outcome you could not know in advance, given what a competent professional in your field already knew. Frame it as a question about a technical outcome — not effort, not a commercial goal — and say why existing methods, prior art or expert knowledge didn't resolve it.HypothesisYour testable, falsifiable proposition about how to resolve the unknown, grounded in established science or engineering. State what you predicted would happen and under what conditions. If it couldn't fail, rewrite it.Experiment(s)The systematic progression of work that tested the hypothesis: what you set up, varied and measured; what you held constant; and why this was a test, not iterative development.Observation & evaluationWhat you actually observed against what you predicted, and the contemporaneous records that captured it (test logs, measurements, datasets, version history, lab notes). Recorded, not asserted after the fact.Logical conclusionWhat you concluded — hypothesis held, rejected, or refined into a new hypothesis for a further cycle. A rejected hypothesis is a valid outcome; new knowledge includes knowing something doesn't work.
1. The technical or scientific unknown
State the specific thing you did not — and could not — know in advance, given what a competent professional in your field already knew. The test is knowledge, not effort. "It was complex and time-consuming" is not an unknown. "Whether approach X could achieve Y under constraint Z, when no established method covered that combination" is.
Frame it as a question about a technical outcome.
Show why the answer wasn't already available (existing methods, prior art, expert knowledge didn't resolve it).
Avoid commercial framing — "we needed it to be cheaper / first to market" describes a business goal, not a technical unknown.
2. The hypothesis
A hypothesis is your testable, falsifiable proposition about how to resolve the unknown — an educated, principled guess you can prove wrong.
R&D hypothesis example: "We hypothesised that a [specific technique] could maintain sub‑[X]ms response times at [N] concurrent operations, where existing [approach] degraded beyond [threshold]."
If your "hypothesis" could not, in principle, fail, it isn't one.
3. The systematic progression of work
Describe the experiment(s) you ran to test the hypothesis, grounded in principles of established science or engineering. This is the part weak narratives skip entirely — they describe building, not testing.
What did you set up, vary, and measure?
What did you hold constant?
Why was this a test of the hypothesis, not just iterative development?
4. Observation and evaluation of results
Record what you actually observed against what you predicted. Reviewers look for evidence that results were captured and assessed — not asserted after the fact. Reference the contemporaneous records: test logs, measurements, datasets, version history, lab notes.
5. Logical conclusions
State what you concluded from the results — whether the hypothesis held, was rejected, or was refined into a new hypothesis for a further cycle. A rejected hypothesis is a perfectly valid (and often more credible) outcome. New knowledge includes knowing that something doesn't work.
Before and after: the same project, two narratives
Weak (describes product development):
"We developed a new logistics platform to give customers real‑time delivery tracking. The integration with third‑party carriers was very challenging and took several months of work by our engineering team. After significant effort we successfully launched the feature on time."
There is no unknown, no hypothesis, no experiment — only a commercial outcome and the assertion that it was hard. Nothing here is assessable against s 355‑25.
Stronger (describes an experiment):
"Unknown: It was not known whether carrier location feeds with inconsistent and intermittent update frequencies could be fused into a single position estimate accurate to within [X] metres, as no established method covered feeds with these reliability characteristics. Hypothesis: We hypothesised that a [named class of] estimation model could maintain [X]‑metre accuracy despite update gaps exceeding [N] seconds. Experiment: We ran controlled trials varying feed frequency and dropout rate against a known ground‑truth dataset, holding the estimation model fixed, and measured positional error. Results & conclusion: Accuracy held below [N] seconds of dropout but degraded beyond it; we rejected the initial model and refined the hypothesis to incorporate [adjustment], which is the basis for the next experiment."
Same project. The second version is assessable — and it is assessable because the experiment was framed before the work, not reconstructed afterward.
Don't forget supporting activities
Most claims also include supporting R&D activities — work that is directly related to a core activity (for example, building a test rig, collecting a dataset, or trialling a prototype to enable the experiment). Describe each one by its direct relationship to the specific core activity it serves. Note that certain supporting activities — broadly, production-related activities and those producing goods or services — must also satisfy a dominant purpose test: they must be undertaken predominantly to support the core R&D, not for another commercial reason (business.gov.au; ITAA 1997 s 355‑30). Self-assess each one; don't assume it qualifies just because it sat near the R&D. A supporting activity cannot save a core activity that does not exist or is not itself eligible.
Where an RSP fits
The recurring failure mode isn't bad science — it's good science described as product delivery, written up at tax time when the experimental framing is gone. As a Registered Research Service Provider, our role is to assist in structuring, documenting, and evidencing R&D activities so that the experimental nature of the work is clearly articulated for R&D Tax Incentive purposes. That is also what turns a contemporaneous record into evidence rather than reconstruction. (For where the line sits, see what does not qualify and our overview of the R&D Tax Incentive.)
Talk to Ignition Research before you register — as a Registered Research Service Provider we help frame the hypothesis and experiment so the narrative holds.
Frequently asked questions
Q: How do I write an R&D activity description? A: Structure it around five elements a reviewer can find: the technical unknown that couldn't be known in advance, your hypothesis, the systematic experiment testing it, your observations against predictions, and your logical conclusions. Describe a test, not a build. Eligibility is self-assessed, so confirm your circumstances with your own adviser.
Q: What is the hypothesis in an R&D claim? A: A testable, falsifiable proposition about how to resolve the technical unknown — an educated guess, grounded in established science, that your experiment could prove wrong. If it couldn't fail, it isn't a hypothesis.
Q: How do I describe technical uncertainty? A: Frame it as a specific question about a technical outcome that a competent professional in your field could not answer in advance from existing knowledge or methods. "It was complex" or "we wanted to be first" don't qualify — those describe effort and commercial goals, not uncertainty.
Q: What makes a core activity description hold up in an ATO review? A: That a reviewer can map it to s 355‑25 without guessing: a genuine unknown, a hypothesis, a systematic experiment, recorded results, and conclusions — backed by contemporaneous records created while the work happened, not written up afterward.
Sources & further reading
Check if you are eligible for the R&DTI (business.gov.au) — core and supporting activities, hypothesis and systematic progression
Income Tax Assessment Act 1997 s 355‑25 — core R&D activities (Federal Register of Legislation)
Research and development tax incentive (ato.gov.au)
Related: Registered Research Service Provider · R&D Tax Incentive overview · What does not qualify
This article is general information from a Registered Research Service Provider about the R&D Tax Incentive. It is not tax, legal or financial advice; eligibility depends on your circumstances and you should self-assess and seek your own advice.
