Quick answer: A technical result can be difficult to interpret without a meaningful reference condition. A baseline or control can help separate the effect of the variable being tested from ordinary variation, environmental change or unrelated system behaviour. Choosing a comparator does not itself make an activity R&D, but where the outcome cannot be known or determined in advance by a competent professional on the basis of current knowledge, information or experience, a well-designed comparison can be important to testing the hypothesis and evaluating what the result means.
A result without a reference can be difficult to explain
Imagine a team modifies an industrial process and records a ten per cent improvement in output.
That sounds promising.
But compared with what?
Was the previous process running under the same load?
Were the input materials equivalent?
Was ambient temperature different?
Did another software setting change between the two runs?
Would the original process also have improved on that day?
Without an appropriate reference condition, the team may know that the result changed without knowing why.
Australian Government guidance describes an experiment as a scientific procedure undertaken to test a hypothesis. It states that records should explain how the experiment was conducted or planned and may, for example, describe the parameters varied, held constant, observed or measured. Evaluation should then consider causal relationships between the parameters varied, held constant, tested or measured.
That makes the comparator part of the experimental logic.
A baseline or control is an experimental-design tool rather than a separate statutory R&DTI requirement; its value depends on whether it helps the experiment test the hypothesis and support a reasoned evaluation.
It is the condition against which the effect of the proposed change can be meaningfully assessed.
Sometimes that is the existing production method.
Sometimes it is an untreated specimen.
Sometimes it is the current software model.
Sometimes the appropriate control is a reference material, a standard configuration or a repeated run in which the experimental variable is deliberately absent.
The right choice depends on the hypothesis.
Start with the effect you are trying to isolate
A useful control helps rule out competing explanations.
Suppose an engineering team is investigating whether a revised cooling method reduces temperature instability in a machine.
The modified system performs better than a run conducted the previous week.
That comparison may be weak if the machine was carrying a different load or the ambient conditions were materially different.
A stronger experiment might operate the existing and modified approaches under comparable loads, measurement methods and environmental conditions.
The research question becomes:
Does changing the cooling method alter the measured response when the other technically relevant conditions are kept sufficiently comparable?
That is more informative than:
Does the new system produce a good result?
This distinction applies across technical fields.
For a materials experiment, the comparator might be the established formulation processed under the same conditions.
For machine learning, it might be the current model evaluated on the same held-out dataset.
For manufacturing, it might be the existing process run against equivalent input material.
For a field trial, the team may need a reference period, untreated area or parallel operating unit.
Ignition Research's existing Insights already use baselines this way in worked examples. The AI contract-review article, for example, discusses defining a comparator before testing, while the AI quoting-engine example compares experimental approaches against an existing rules-based method on a defined holdout.
The important point is not that every project needs the same kind of control.
It is that the comparison should answer the technical question rather than merely make the result look favourable.
The easiest baseline is not always the right one
Several common choices can produce misleading comparisons.
The first is the historical average.
Historical performance can be useful, but conditions may have changed. Equipment ages, operators change, material batches vary and software versions move.
If those changes influence the outcome, the historical number may not represent the condition the experiment needs to compare against.
The second is the best previous result.
This can make a useful benchmark for performance, but it may be a poor experimental control if that result came from different conditions.
The third is the manufacturer's published specification.
A specification can tell you what a commercial product is designed to achieve. It does not necessarily tell you how your modified system would have behaved under the same conditions as the experiment.
The fourth is no-change operation during the trial.
This can be stronger because it produces a contemporaneous comparison, but only if the control and experimental condition are exposed to sufficiently similar inputs.
A baseline should therefore be judged by its ability to rule out competing explanations.
This is a practical experimental-design framework rather than a prescribed R&DTI template. A pre-test note might ask:
• What would happen if the experimental change were not made?
• Can that condition be observed directly?
• Which factors must be comparable between the baseline and experimental run?
• Which unavoidable differences need to be recorded?
• Is the measurement method the same?
• Are both conditions exposed to the same input population or operating envelope?
• Could the baseline itself drift during the experiment?
• What difference between the two would be technically meaningful?
These questions do not guarantee a perfect control.
Commercial R&D rarely happens in perfectly controlled environments.
They do, however, make the limitations of the comparison visible.
Sometimes the baseline has to change
A project can also outgrow its original comparator.
Early in development, the baseline may be an existing internal process.
Later, that process may no longer be the relevant reference.
A new industry method may become available.
The product architecture may change.
The research question may move from “can this mechanism work?” to “can it outperform the established alternative under the required operating conditions?”
Changing the comparator is not automatically a problem.
Changing it silently is.
If the baseline changes, the research record should preserve:
• the original reference condition;
• why it was chosen;
• what evidence showed it was no longer adequate;
• the new comparator;
• why the new comparator better matches the revised question; and
• whether earlier and later experimental results can still be compared directly.
This matters because otherwise the project can appear to improve simply because the reference moved.
Current R&DTI record-keeping guidance states that records should be created at the time the activity is conducted. Examples of records that may demonstrate a systematic progression of work include experimental plans, records of trial runs and test reports showing results and evaluation.
That principle applies directly to changing controls.
A baseline is part of the experimental design.
If it changes, the reason is part of the experimental history.
A good comparison should inform the next decision
The purpose of a baseline is not to make a chart look more scientific.
It is to help the team make a technical decision.
Suppose the experimental system performs no better than the control.
That may weaken the hypothesis.
Suppose it performs better only under one operating condition.
That may reveal a narrower operating envelope.
Suppose both the baseline and experimental system improve at the same time.
That may suggest an external variable rather than the proposed change caused the effect.
And suppose the experimental system consistently outperforms the reference under controlled comparisons.
That may justify testing robustness, scale or field conditions next.
Each of these outcomes can change or sharpen what the team knows. That is what a meaningful comparator should help achieve.
That is what a meaningful comparator should do.
The strongest question before running the experiment is therefore not:
“What number should we beat?”
It is:
“What comparison would allow us to tell whether the proposed change actually caused the effect we care about?”
If the team cannot answer that, collecting more data may not solve the interpretation problem.
Conclusion
A baseline or control does not make an experiment scientific simply by being present. Its value lies in whether it creates a meaningful comparison for the hypothesis being tested.
The strongest comparator is one that keeps relevant conditions sufficiently comparable, exposes competing explanations and makes clear what difference would change the technical interpretation.
When the reference condition changes, that change should remain visible in the research record so that later results can still be understood in context.
Talk to Ignition Research if your team has a hypothesis and test plan but the appropriate control, baseline or reference condition is unclear. As a Registered Research Service Provider, Ignition Research assists businesses with research question definition, experimental design, technical investigations and contemporaneous supporting records within its registered research fields.
RSP registration confirms that the provider has met the relevant registration requirements for its registered research fields. It does not constitute government endorsement of a project or guarantee R&DTI eligibility.
We do not determine R&DTI eligibility or provide tax advice. The R&D Tax Incentive is a self-assessment program, and companies remain responsible for assessing their activities, expenditure and circumstances and obtaining appropriate tax advice.
Sources & Further Reading
Australian Government — Conducting core R&D activities for the R&D Tax Incentive
Australian Government — Record keeping for the R&D Tax Incentive
R&DTI note: A baseline or control does not itself determine R&DTI eligibility. Its role is to support the experimental design and evaluation; eligibility still depends on whether the activity satisfies the applicable R&DTI requirements.
Working on a project like this?
Tell us what you're building and the part you're not sure will work. A researcher from our Adelaide team (registered Research Service Provider RSP000047) reads every enquiry and replies within two business days. No documents needed.
