Incrementality testing – experimentation that measures the true incremental contribution of a media channel, campaign, ad set, or tactic to business results – is critical for savvy advertisers who recognize the myth of marketing measurement: the false belief that ‘advertising measurement is inherently defined by total, deterministic precision.’
Eric Seufert, in his podcast, suggests that platforms relying on identity resolution as a single source of truth for campaign measurement reveal more about the flawed mindset of the platform than the actual performance of ad campaigns. In a post-ATT world on the cusp of a cookieless future, deterministic tracking is likely not a broadly reliable input for proving ad performance at the level of placement, channel, or strategy.
Fortunately, privacy-safe, open-source, “inwards-focused” statistical models such as causal impact offer a reliable alternative. These models are built on the principle that well-powered experiments and statistical inference can establish a baseline for business impact, against which actual performance can be compared. This approach provides actionable insights without resorting to weak attempts to “map” the customer journey across devices and browsers.
The challenge for those controlling advertiser budgets is that experimentation, particularly randomized control trials (RCTs), the gold standard, is both expensive and time-consuming. The costs include direct ad spend as well as the opportunity cost of “turning off” markets or audiences to create control groups. Additionally, setting up well-powered tests requires significant effort.
Opportunities for Improvement
Two key players in the ecosystem can address these challenges:
For ad networks:
- Reduce the cost of launching campaigns against a market or audience. Lowering these costs enables greater variance between markets, making it easier to isolate and measure ad campaign impacts.
- Reduce the time required to launch an ad on a given inventory. Faster activation improves the likelihood of running successful experiments by ensuring ads appear precisely when and where they’re needed.
For measurement providers (platforms/partners):
- Implement automated incrementality testing to reduce scale requirements. Instead of shutting off an entire country (e.g., the U.S.) to test a marketing intervention—an easy but risky approach—advertisers can run multiple, smaller-scale tests across different markets using automated tools.
- Adopt improved Media Mix Modeling (MMM). Advanced Bayesian MMM, which incorporates prior sales data and KPIs to estimate posteriors, allows for more precise forecasting and stronger causal inference. This approach not only determines each channel’s contribution to overall performance, but also estimates the effectiveness of advertising relative to prior expectations.
At AdQuick, we partner with companies that implement all forms of MMM effectively. Additionally, we help reduce both the cost and time required to launch out-of-home (OOH) campaigns. Learn more on our website.
