Decision-Making characteristics framework for marketing attribution in practice Improving empirical procedures

In today's multi-channel environment, It is becoming increasingly difficult to implement advanced multi-touch attribution (MTA) models to facilitate advertising decision-making. This is due to the rising number of advertising platforms, such as TikTok, Metaverse and Google -- each with its own unique attribution principles -- and the decline in user-level disaggregated data. Over time, the development of marketing models has matured in parallel with the greater availability of consumer data and understanding of consumer behaviour. To overcome media optimisation challenges at the tactical and channel levels, e-commerce brands have replaced traditional media-mix methods with attribution methods that provide immediate insights into return on advertising spend. However, existing MTA models lack simplicity, robustness, ease of interpretation, and accuracy, all of which are critical attributes of decision-supporting models. To address this, this paper proposes a holistic conceptual framework that captures the various interplaying characteristics of attribution models in practice. The concept is based on the evolution of modelling and insights into the development of decision-making paradigms. The proposed architecture highlights the interactions between various tools, media categorisation and metrics, and how they influence media spend optimisation at the channel and tactical levels. The paper also describes some of the most recent advances in media measurement practices. By adopting the proposed framework, future advertisers can identify the best way to overcome the challenges of analysing marketing performance.