Incrementality testing has continued to mature as a measurement practice through 2025 and into 2026. The pattern at the larger Australian advertisers is now to run rigorous incrementality tests on the major channels at a regular cadence rather than to treat incrementality work as a one-off project. The methodologies have stabilised, the operational discipline has improved, and the results are being more reliably integrated into planning decisions.

Where the practice has landed in May 2026:

Geo-holdout tests are the dominant methodology for channels where they are feasible. The pattern is to define a set of test geographies and a set of control geographies, hold spend off in the test geographies for a period, and compare outcomes against the control. The methodology has matured to handle the operational complications — synthetic control construction, geographic spillover, and exogenous shocks.

Audience-based holdout tests are used where geo-holdouts are not appropriate. The pattern is to define a holdout cohort within an addressable audience, suppress advertising delivery to that cohort, and compare outcomes against the exposed audience. The methodology requires identity continuity through the test period and is operationally more demanding than geo-holdout.

Conversion-lift studies on platforms that support them have become routine for paid social and display work. The Meta, Google, and TikTok conversion lift study tooling has matured significantly and is being run routinely by advertisers operating at scale on those platforms.

Switch-back tests are used selectively. The pattern of running a campaign on for a period, then off, then on, is operationally simpler than holdout designs but is more sensitive to underlying trend changes and seasonal effects. It works where the underlying trend is reasonably stable but is harder to interpret where it is not.

What is working operationally:

Pre-registration of test design. The advertisers running rigorous incrementality work are pre-registering the test design — the hypothesis, the methodology, the analysis approach, the success criteria — before the test starts. The discipline reduces post-hoc analysis drift and supports more credible results.

Sample size planning. The test designs with explicit power analysis at planning stage are typically delivering credible answers. The test designs without explicit power planning are too often reaching the end of the test period with results that are not statistically distinguishable from zero.

Trend control. The serious incrementality work uses synthetic control or formal differences-in-differences methods to control for underlying trend rather than relying on simple before-versus-after comparisons. The methodology choice matters for credibility.

Multi-test cadence. The advertisers integrating incrementality testing into planning are running tests on a recurring cadence — typically quarterly or semi-annually on the major channels — rather than running one-off tests. The recurring cadence builds a knowledge base over time that supports planning decisions more reliably than individual test results.

What remains difficult:

Channel interactions. Testing the incremental value of one channel while holding others constant is operationally clean. Testing the interactions between channels — where channel A increases the effectiveness of channel B — is much harder. The advertiser-level evidence on channel interactions is thin in 2026 and the analytical methods for measuring it are immature.

Long conversion windows. Channels and campaigns aimed at longer conversion cycles — B2B, considered consumer purchases, long-cycle financial services — are harder to test through incrementality methods because the test period needs to be long enough to capture the conversion outcomes. The patience required for these tests is sometimes greater than the operational decision cycle supports.

Brand effects. Brand-led channels — sponsorship, brand video, large-format outdoor — have effects that are harder to attribute and harder to test through incrementality methods. The available methodologies for testing brand effects rigorously are operationally heavier than for performance channels and the long-run sample sizes required are larger.

Cross-platform identity. Where the test depends on identifying exposed-versus-control individuals across platforms, the identity-continuity challenge of 2026 makes the work harder than it was in 2018. The methodologies are adapting but the friction is real.

Operational notes for advertisers planning incrementality work through the rest of 2026:

Annual measurement plan. The advertisers getting the most value from incrementality testing are running an annual measurement plan that schedules specific tests on specific channels at specific times. The discipline produces a coherent knowledge base over time.

Decision integration. The tests that change planning decisions are the tests where the planning team is involved in test design and the test outcomes are presented in formats that map directly to planning decisions. Tests that produce academically interesting outcomes that do not connect to planning decisions tend not to change behaviour.

Realistic statistical expectations. Many incrementality tests do not produce clean point estimates of incremental return. They produce ranges with confidence intervals that often span a wide band. The decisions based on the test results need to account for this uncertainty rather than treating point estimates as fact.

Documentation of methodological choices. The methodological choices in incrementality test design have meaningful effects on the result. Documenting the choices supports credibility and supports the long-run value of the test results across multiple stakeholders.

For Australian advertisers and measurement teams in mid-2026, the working read is that incrementality testing is a mature, well-understood part of the measurement toolkit. The methodologies are reliable when applied carefully. The operational discipline of recurring testing builds a knowledge base that supports better planning decisions. And the work is meaningfully more rigorous than the attribution-only measurement that was standard practice five years ago.

The next 12 months will likely bring more cross-channel testing methodology development, more rigorous testing of brand effects, and continued operational maturation of the testing cadence across more Australian advertisers. The practice is healthier than it has been at most points in the last decade.