TL;DR:
Businesses aim for clear cause-and-effect relationships to inform their decisions, aiming to maximize returns while minimizing risks and uncertainty. However, true causality is impossible to achieve. As a solution, incrementality allows businesses to assess the impact of their decisions by sequentially measuring how changes in actions, such as advertising spend, affect outcomes. This method helps businesses refine strategies, identify consequential levers, and make more data-driven decisions.
One common way to reason how your actions will impact the outcome of interest is to evaluate through the lens of causal inference. Most extensively studied in economics and natural science literature, the concept and method of building the causal relationship is impossible to accomplish (See Andrew Gelmen’s comment on Nobel prize in economics for causal inference).
For example, let’s say we want to quantify the impact of paid media advertising investments on beer sales during the Super Bowl. We would need to isolate out the baseline sale and event-specific demand, as well as the interactions between the factors (see the diagram below). The approach (technically referred to as ‘identification strategy’) can be subjective, and the resulting insights may not be practical enough for businesses to act upon.
To navigate the complexities and avoid fallacies, we can instead infer the impact by comparing outcomes from controllable actions that are intentionally and repeatedly adjusted. In marketing and business analytics practice, this approach is called ‘incrementality.’
- Causality: I expect B to happen if I do A
- Incrementality: I expect B to change by y if I change A by x
Differing from coincidence or correlation, the incrementality measurement can isolate and quantify how decisions drive outcomes. So, how do we strategically plan and execute to properly measure incrementality?
What does Measuring Incrementality in Marketing Look Like?
(1) Define Levers & Test Strategies
Start by defining key decision levers that can be controlled and actionable, such as investment (ad spend) levels, campaign strategy, different targets (markets and segments), and timing. Combining these levers gives us test strategies to plan how each lever will be evaluated. A wider range of test strategies leads to a more accurate assessment of what truly drives performance.
For a simple strategy, consider testing the impact of different ad spend levels as a lever. A test strategy can be defined in three different strategies: testing in low, moderate, and high budget levels. By measuring sales across these groups, the business can determine the optimal spending level that maximizes returns.
For a more complex strategy, one can combine different levers: ad spend levels, campaign messaging, and regional targeting. For example, a multi-city experiment could compare a brand-awareness campaign in one region against a direct-response campaign in another, all while varying budget allocations. This approach helps understand not only which strategy drives higher returns but also how effectiveness varies across different markets.
(2) Actions: Collect Data
For implementations, simultaneous controlled trials can be leveraged, where different strategies are rolled out at the same time. The outcomes of each strategy can be compared side by side with an assumption that other factors have equally impacted the strategies (e.g., A/B testing). Another method is before/after comparisons, where you compare results before and after implementing the strategies. A hybrid approach, combining both methods, can also provide data to measure the incremental changes when the strategies are more nested or complex.
(3) Measurement: From Data to Insights
To evaluate the tests, different statistical approaches can be used depending on the scope and data availability. From the analysis, the incrementality can be quantified as a change in return relative to the change in advertisement spendings. Unlike Return on Ad Spend (RoAS), which measures the average return on investment of all historical period (RoAS = Return/Investment), incrementality focuses on the marginal return—the additional return from the controlled changes in investment levels (mRoAS = ΔReturn / ΔInvestment).
By analyzing multiple estimated mRoAS (i.e. diminishing return curves), decision makers can identify the point where additional spending stops delivering intended and proportional gains. This insight enables smarter budget decisions—scaling up when returns are strong and reallocating funds when efficiency declines.
(4) Repeat
Incrementality measurement is an ongoing process that requires continuous implementations and reinforcements. Market conditions, customer behaviors, and the competitive landscape continuously change. To adapt to these uncontrollable shifts, ongoing incrementality measurements can help track how the effectiveness of different levers changes in real time.
Incrementality Is A Part of Everything
Leveraging incrementality empowers businesses to make smarter, data-driven decisions by pinpointing which and how marketing actions genuinely drive growth. Still, like any measurements, there are challenges that need to be considered when taking actions. In the next article, we’ll explore what considerations and risks involved in measurements and how practitioners can handle delivering more robust insights to stakeholders.
Building with Incrementality
A culture of reasoning ensures decisions are proactive, not reactive, based on continuous and intentional insights. We can guide you through how incrementality can be part of the culture.
