Marketing attribution models identify which advertisements were seen or clicked before conversions, but they often mis-state true impact. Incrementality measurement, usually via experiments, reveals the causal lift from marketing — the conversions that would not have happened without the advertisements. This paper bridges the gap between attribution and incrementality by using attribution data to predict incremental outcomes. It outlines a regression-based modelling approach that learns from instances where incremental lift has been measured (through lift tests or natural experiments) and uses attribution signals to estimate lift for other campaigns. Key challenges in scaling incrementality testing (high cost, delays and complexity) are addressed by leveraging abundant attribution data intelligently. The paper presents practical techniques and best practices for building incrementality prediction models, from selecting features (eg conversion propensities, user touch points) to validating predictions with ongoing experiments. Real-world examples in e-commerce and mobile apps illustrate how these models inform budget allocation, campaign optimisation, channel mix, targeting and executive decision-making. Findings show that combining attribution’s granularity with incrementality’s truth can provide near-real-time guidance on where marketing dollars truly add value. The implications for practitioners are significant: embracing an incrementality mindset enables more objective marketing performance evaluation, fosters cross-functional trust (especially with finance) and leads to more efficient, evidence-based marketing investments. This work provides a roadmap for marketers to evolve from merely counting conversions to reliably quantifying marketing’s incremental contribution to business growth. This article is also included in The Business & Management Collection which can be accessed at http://hstalks/business/.
Seojoon Oh (Sun,) studied this question.