Quick answer: More data does not automatically make an experiment more informative. Additional observations strengthen the research where they help test the hypothesis, distinguish between explanations or reduce relevant uncertainty. Large volumes of monitoring data can still leave the original technical question unanswered if the relevant comparison and evaluation approach were not defined clearly enough to test the hypothesis.
The 2026–27 Federal Budget announced R&DTI reforms for income years starting on or after 1 July 2028. Until those changes take effect, the R&DTI continues to be administered under the current legislation.
Data volume and experimental evidence are different things
Modern technical projects can collect enormous amounts of data almost by default.
A field device may log every few seconds. A manufacturing line may retain sensor histories. A software platform can capture latency, errors, user events and system states continuously.
It is tempting to assume that more observations must produce stronger evidence.
But an experiment is not defined by the amount of data collected.
It is defined by what the data allows the team to conclude about a hypothesis.
Australian Government guidance describes observation as one part of a wider systematic progression that begins with a hypothesis and experiment and continues through evaluation to logical conclusions. Collecting observations is therefore not, by itself, the experimental method; the observations need to be capable of informing the evaluation of what was tested.
That sequence matters.
If a team collects thousands of measurements but never defined what comparison those measurements were meant to support, the dataset may be operationally valuable without resolving the research question.
Start with the decision, then work backwards to the data
Suppose a team is testing whether a new control method reduces instability in a renewable-energy system under variable operating conditions.
It could collect:
• weather conditions;
• voltage;
• current;
• equipment temperature;
• vibration;
• power output;
• control commands;
• error states;
• communications latency; and
• dozens of derived metrics.
All may be technically interesting.
But which ones are needed to answer the hypothesis?
That depends on the technical explanation or relationship the hypothesis proposes.
If the hypothesis concerns how the control method responds to rapid input variability, the experiment needs measurements that capture that relationship.
If the hypothesis concerns thermal instability, a different measurement set may be required.
The data plan should therefore follow the research logic:
What result are we trying to explain?
What variable do we believe causes or changes that result?
What comparison would support or weaken that explanation?
What measurements are required to make that comparison?
Only then does collection volume become meaningful.
Extra data can create new problems
Collecting more variables can sometimes make interpretation harder.
The team may find patterns after the experiment that were never part of the original hypothesis. Some may be real. Others may result from chance, confounding variables or differences in operating conditions.
That does not mean unexpected observations should be ignored.
They can be valuable sources of new hypotheses.
But a distinction should be maintained between:
evidence used to test the planned hypothesis, and
observations that suggest a new question for later investigation.
Without that distinction, a team can unintentionally rewrite the purpose of an experiment after seeing the data.
The government guidance on systematic progression expects the hypothesis to precede the experimental work and the evaluation to analyse what the results mean in relation to what was tested.
An unexpected correlation may therefore provide a basis for a new hypothesis to be tested in a subsequent experiment, rather than being treated immediately as a conclusion about the original hypothesis.
Record why a dataset exists
One practical improvement is to give each major dataset a research purpose.This is a practical record-keeping discipline rather than a prescribed R&DTI template.
A record can identify:
• the experiment it belongs to;
• the hypothesis being tested;
• the source of the data;
• the variables used in the planned analysis;
• data excluded and why;
• transformations applied;
• version or time period;
• known limitations; and
• the conclusion the dataset is capable of supporting.
This becomes especially important when operational systems continuously overwrite or update information.
Ignition Research's existing work on asset analytics already notes that reproducible data snapshots and version records can matter when experimental datasets sit inside changing operational pipelines.
The underlying principle extends beyond software.
If a result cannot later be tied back to the data, conditions and version that produced it, the evidentiary value of having collected more information may be limited.
Know when more experimental data is unlikely to change the decision
There is also a management question.
Teams sometimes continue collecting observations because stopping feels premature.
But every research cycle should eventually reach a point where the evidence is sufficient to make the decision it was designed to support.
That decision might be:
• revise the hypothesis;
• run a more targeted experiment;
• reject one technical approach;
• proceed to a pilot;
• define an operating limit; or
• stop the research line.
More data is most useful when it can materially change the decision, reduce decision-relevant uncertainty or strengthen the basis for the conclusion.
If additional experimental monitoring is unlikely to change the conclusion or materially reduce the remaining uncertainty, the next step may no longer be further data collection.
It may be evaluation, redesign or implementation.
More data and R&DTI eligibility are separate questions
The volume of data collected does not determine whether an activity is a core R&D activity. The relevant questions remain whether the outcome could be known or determined in advance, whether it could only be determined through a systematic progression of work, and whether the activity was conducted for the purpose of generating new knowledge. Routine monitoring, data collection or analysis may generate useful information without, by itself, satisfying those requirements.
Conclusion
More data can improve an experiment, but only when the additional observations help resolve the question the experiment was designed to answer.
A useful data plan begins with the hypothesis and the decision the evidence needs to support, then identifies the measurements, comparisons and context required to evaluate that hypothesis.
When additional observations no longer materially reduce the relevant uncertainty, the research team should decide whether the next step is a more targeted experiment, a revised hypothesis or a move into implementation.
Sources & Further Reading
Talk to Ignition Research if a technically uncertain project is generating substantial data but the link between the measurements, hypothesis, evaluation and research decision 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.
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.
