Why incrementality matters
In today’s mobile advertising landscape, measuring marketing performance has become increasingly complex. User acquisition campaigns take place in environments characterized by a multitude of aspects such as seasonality, algorithms, app store visibility, competitor activity and overlapping targeting, which makes it even more difficult to understand which efforts are responsible for incremental growth.
Traditional metrics, such as clicks and impressions, and attribution models tend to oversimplify both reality and the customer journey, failing to identify the real forces behind positive UA growth outcomes. Marketers are shifting towards more aggregated and privacy-based methodologies, and this is where the concept of incrementality gains prominence: 71% of advertisers consider it the most important KPI in retail media networks and in the user acquisition field.
What is incrementality
Incrementality refers to the measurement of the true causal impact of a marketing activity, meaning the extent to which a user acquisition campaign generates outcomes that wouldn’t have occurred otherwise.
Unlike traditional performance metrics, incrementality focuses on causation instead of correlation, isolating cause and effect. In standard reporting, the emphasis is on a relationship: an user sees an ad, interacts with it and then converts. Despite the connection, this does not prove that the marketing activity actually caused the outcome, since it might have happened for other reasons, such as previous intent, organical conversion or external factors. Incrementality challenges this assumption by asking whether the UA campaign has had a real impact on the user behavior or if it has simply converted the already existing demand.
To quantify this, incrementality introduces the concept of lift, which is the additional outcomes generated by a marketing activity. Instead of measuring revenues or conversions, incrementality isolates the incremental gains, meaning the new values attributable to the user acquisition campaign. This makes it possible to identify the activities that have positive consequences on growth and revenue goals and those who don’t, to eventually intervene and change them.
Measuring incrementality requires a structured approach, creating a comparison between exposed and unexposed groups, allowing marketers to approximate what would have happened otherwise. As a result, incrementality is inherently tied to experimentation and testing methodologies, which are essential for accurately estimating causal effects within the user acquisition campaign.
Incrementality vs Attribution models
Although the terms are sometimes confused, attribution and incrementality are two different concepts. Attribution is the process of matching two touchpoints of data: whether it’s last-click, first-click or multi-touch attribution, the aim is to distribute value, such as installs or in-app events, across channels. Despite that, attribution does not equal causation.
Incrementality, on the other hand, quantifies the true effectiveness of an app’s marketing activity, it asks if a specific channel actually made a difference instead of who receives the credit. Incrementality does not replace attribution, but gives it a more robust layer of insight.
The two can be in fact used in conjunction with media mix modeling (MMM): attribution allows marketers to understand how to optimize UA campaigns in the short-term; with incrementality they are able to test new channels and markets in the mid-term, while MMM is a perfect pairing for long-term planning since it allows them to forecast the best allocation of their budgets to optimize their user acquisition campaign. Regarding this aspect, Appsflyer offers a tool that enables advertisers to monitor their user acquisition campaign thanks to a fully automated causal measurement engine. In their dashboard, incremental results will appear alongside key attribution metrics, to enable faster optimizations and immediate budget shifts.
How to measure incrementality
Measuring incrementality requires a shift from tracking to testing. Traditionally, it involved control groups and test groups, which were respectively not exposed and exposed to the user acquisition campaign. By comparing the outcomes between these groups, marketers can determine the incremental lift generated by the UA campaign. This method brings numerous benefits, such as helping understand what works and what doesn’t, and can be a massive advantage and key element in the success of a UA campaign.
Despite its effectiveness, this method can be expensive and time-consuming to perform, and it’s becoming even more difficult to implement the level of detail needed for incrementality A/B tests due to privacy regulations. Consequently, causal inference and other methods have become relevant approaches for quantifying incrementality:
Causal inference uses statistical techniques to isolate the causal impact by comparing treated and control groups through synthetic control groups or other models. It provides clear and accurate causal relationships and also takes into consideration external factors, but it can be costly and difficult to implement.
Budget holdout consists in a portion of the marketing budget withheld from certain regions or groups to act as a control, while the rest receive the full user acquisition campaign. It allows a clear understanding of the impact of increased or decreased spend but requires significant resources and there may be interferences with other concurrent campaigns
Geo-lifting, instead, involves dividing geographical regions into test and control groups to measure the impact of marketing efforts in specific areas. It can be implemented quickly, but it requires significant resource investment and detailed knowledge, as well as being less accurate in isolating the true effect of marketing.
In the last few decades, AI has grown more and more important in the field of incrementality and user acquisition: it guarantees more speed, accuracy and scalability than traditional methods, it helps identify causal impact more precisely through better test/control matching and more advanced MMM. AI also enables synthetic control creation when true comparisons are not possible and supports optimization based on incremental performance. In the end, AI allows DSPs to run continuous and automated A/B tests to learn and scale effective tactics without manual intervention.
Challenges and limitations
Despite the benefits deriving from incrementality in measuring marketing effectiveness, implementing it is complex. One of the main challenges is designing and executing reliable experiments, since it requires not only technical infrastructure but also understanding of statistical principles to ensure that results are valid and interpretable. Poorly designed experiments can lead to misleading conclusions.
Another significant barrier is organizational, since channels that appear effective in traditional reports may show limited incremental impact when tested, while data limitations also play a crucial role, because running precise experiments is challenging due to the fact that an accurate incrementality measurement depends on the ability to track and compare user behaviors in a privacy-first environment.
In addition, incrementality might encounter other difficulties, such as the fact that what works in a controlled test setting might not translate perfectly in real-world conditions relating to user acquisition, or the fact that lift is rarely uniform across audiences, time periods or channels and results can vary due to external factors such as seasonality or competitive activity.
Why Mapendo is the answer to incrementality
Mapendo is a user acquisition platform designed to achieve your incrementality goals from day-1, with a zero-hassle MMP integration. Our performance-based solution allows advertisers to pay only for valuable results, a.k.a. users who are going to generate in-app revenue. We don’t replace your current partners, but we integrate with them, providing you with a global and premium inventory, across 50+ diversified channels (programmatic, rewarded, performance, CTV, etc.).
You don’t have to choose between scale and profitability anymore. Mapendo combines its historical data in your vertical with regular testing of new placements, in order to reach the best audiences for your app. By leveraging AI and proprietary ML algorithms, we optimize every channel towards your post-install goals and KPIs, improving performance over time.
