Between 30 and 70 per cent of conversions attributed to digital campaigns would have occurred without the advertising. Correlation-based measurement is no longer a defensible governance standard for organisations making material marketing investments.
The Correlation Problem at the Heart of Marketing Measurement
For the better part of two decades, digital marketing measurement has been built on a foundational assumption that has never been rigorously tested in most organisations: that correlation between advertising exposure and conversion constitutes evidence of causation. The logic seems intuitive — a customer saw an ad, and then they purchased. The ad worked. But this inference is almost always wrong in ways that are material to budget decisions, and the accumulated cost of that error is significant.
The core problem is selection bias. Digital advertising systems — particularly retargeting and search — are designed to reach people who are already likely to convert. An individual who has visited a product page, added an item to a cart, and searched for a brand by name has already demonstrated high purchase intent. Showing them an ad at that moment and attributing the subsequent conversion to the ad is not measurement. It is post-hoc rationalisation dressed in the language of data.
The incremental question — would this conversion have occurred without the advertising intervention — is the only question that matters for understanding whether a marketing investment has generated value. Yet it is the question that almost no standard analytics implementation answers. The cost of this omission is not trivial. Analyses of incrementality testing results consistently find that between 30 and 70 per cent of conversions attributed to digital campaigns would have occurred regardless of the advertising. The specific figure varies by channel, industry, and targeting approach — but the direction of the bias is consistent.
What Incrementality Testing Actually Measures
Incrementality testing addresses the correlation problem through controlled experimentation. By dividing an audience into exposed and unexposed groups — where the unexposed group is withheld from advertising — and measuring the difference in conversion rates between the two groups, it becomes possible to isolate the causal contribution of the advertising intervention. The incremental conversion rate is the difference between observed conversions in the exposed group and the counterfactual conversion rate estimated from the control group.
This methodology produces what is genuinely missing from standard attribution: an estimate of what would have happened without the advertising. The counterfactual is not inferred from attribution models, assumed from historical benchmarks, or borrowed from industry studies. It is observed directly from a real control group experiencing the real commercial environment at the same time as the exposed group. This is the gold standard of causal inference, and it is what separates incrementality measurement from correlation-based attribution.
Between 30 and 70 per cent of conversions attributed to digital campaigns would have occurred without the advertising. The direction of the bias is consistent across channels and industries.
Major platforms — including Meta, Google, and The Trade Desk — now offer varying forms of geo-based or user-based lift testing that approximate this methodology. But it is important to distinguish between platform-administered incrementality tests and independent incrementality measurement. Platform tests are conducted within the platform’s ecosystem, using the platform’s measurement infrastructure, in a context where the platform has an economic interest in a particular outcome. Independent testing — using third-party measurement providers or in-house experimental design — is a different standard of evidence.
The Practical Challenges of Running Incrementality Tests at Scale
Incrementality testing is not a simple operational task. It requires withholding advertising from a statistically meaningful control group, which creates friction with channel managers and media agencies whose performance metrics are calculated on the exposed population only. It requires holding other variables constant during the test period — pricing, promotional activity, competitive conditions — which limits the windows in which valid tests can be conducted. And it requires sufficient conversion volume in both groups to produce statistically significant results, which restricts its practical application to higher-volume channels and campaigns.
Why the Industry Has Been Slow to Adopt Rigorous Standards
The adoption of incrementality testing has been slower than the evidence base would suggest it should be, and the reasons are instructive. The primary barrier is not technical — the methodology is well understood and the tools are available. The barrier is the risk that rigorous testing will reveal that significant portions of current advertising spend are generating minimal incremental return. This is not a hypothetical concern. Organisations that conduct their first systematic incrementality tests frequently discover that some of their highest-spend channels are delivering the lowest incremental return, precisely because those channels are most effective at reaching people who were already going to convert.
The political economy of this discovery is uncomfortable. Channel managers, media agencies, and technology vendors all have economic interests in the current measurement paradigm. Attribution systems that show high returns justify continued investment and generate agency fees. Incrementality testing that reveals low returns creates pressure for reallocation that disrupts existing commercial relationships. The measurement reform is opposed, subtly or explicitly, by the constituencies most exposed to its findings.
The Governance Case for Mandatory Incrementality Standards
The board-level implication of the incrementality imperative is straightforward: correlation-based measurement is no longer a defensible governance standard for organisations making material marketing investments. In a regulatory and commercial environment that increasingly demands rigorous justification of capital deployment, the ability to demonstrate that marketing expenditure generates incremental business outcomes — not merely correlated outcomes — is both a commercial and a governance requirement.
Progressive Australian organisations are beginning to embed incrementality testing requirements into agency contracts and media briefs, requiring that a specified percentage of annual spend be subject to controlled testing on a rolling basis. This approach does not eliminate attribution-based reporting — which remains useful for operational optimisation — but it establishes incrementality measurement as the authoritative standard for strategic budget decisions. The organisations that move earliest to this standard will have a durable advantage in capital efficiency over those that continue to defend spending on the basis of correlation.