Advertising is in a measurement transition. The platform-attributed conversions that dominated the last decade are increasingly understood as overstated. The third-party cookie that powered cross-site tracking is largely gone. Privacy regulation has tightened the data available to attribution platforms. The combined effect is that an advertising strategy built around platform-reported ROAS now sits on shakier ground than it did three years ago. The strategies that hold up are built with the new measurement reality in mind from the start.
The Three Measurement Approaches That Matter Now
Incrementality testing — running geographic or audience-based hold-out experiments to measure the actual lift attributable to advertising. Media mix modelling (MMM) — statistical models relating spend across channels to outcomes, validated against incrementality tests. Platform attribution — still useful for in-channel optimisation, but treated as one input among several rather than the source of truth. The mature measurement stack uses all three, with incrementality as the gold standard, MMM as the always-on model, and platform attribution as the operational signal.
Designing Campaigns for Measurement
A campaign designed for measurement looks different from a campaign designed only for performance. It is large enough that lift can be detected against background variability. It runs long enough that the test has statistical power. It includes hold-out groups built into the design rather than reverse-engineered from observational data. It treats branded search as a separate measurement question, not as conversion attribution. None of this changes the creative; it changes how the campaign is structured and what data is available at the end.
The Brand vs Performance False Dichotomy
A common framing pits brand spend against performance spend, with finance teams expecting performance to justify itself in the next quarter and brand to justify itself eventually. The framing is increasingly unhelpful. Brand spend that builds memory structures has measurable performance effects — incrementality tests can demonstrate the lift, MMM can attribute the contribution, and the time horizon for measurement is six to twelve months rather than indefinite. Treating the spend categories as separable produces over-investment in last-click performance and under-investment in the demand-generation work that performance depends on.
An advertising leader's most consequential ongoing question: what proportion of conversions you are seeing through performance channels would have happened anyway? The honest answer for most categories is meaningfully more than zero. The implication is that aggressive performance optimisation against platform-reported conversions tends to over-allocate to lower-funnel work that captures existing demand and under-allocate to upper-funnel work that creates demand. Incrementality testing is the only way to surface this distinction credibly.
The Channels That Are Changing
- Connected TV — measurable in ways traditional broadcast was not, with pricing still finding equilibrium
- Retail media — first-party data networks from large retailers, increasingly important for CPG
- Influencer at scale — moved from experimental to mainstream, with measurement still maturing
- Search — generative AI is reshaping query patterns; planning needs to follow user behaviour
- Programmatic display — privacy changes have reduced effectiveness; reweighting toward contextual
- Out-of-home — digital OOH with measurement integration is no longer the unmeasurable channel it once was
What Holds Up Across the Transition
The fundamentals are stable. Distinctive brand assets that build memory structures over time. Reach across the relevant audience rather than narrow targeting on lookalike heavy users. Creative that earns attention and is fluently recognisable as the brand. Long-term brand investment alongside short-term activation. These principles predate the platform era and will outlast its current measurement crisis. Strategies built on them are robust against measurement transitions; strategies built on transient platform mechanics are not.