Actionable Analytics: Metrics, Models, and Smarter Marketing
(A2025-124940)

AUTHORS

RIK PIETERS, Tilburg University - Catolica University Lisbon

ABSTRACT

New data, metrics, methods, models, and machine learning procedures become available almost at the speed of light. This offers unbounded opportunities for marketing analytics to improving research practice, and the welfare of consumers, business and society. However, these developments also hold threats of overenthusiasm, drowning by data, and diving into piecemeal, scattered research and marketing projects that fail to live up to their promises. The computer technology company Oracle identifies data quality, access and visualization as among the top three priorities of data analytics (https://www.oracle.com/eg/business-analytics/data-analytics-challenges/#top-10, June 27, 2024). Early on, marketing experts (Mela and Moorman 2018) identified the challenges for big data and data analytics initiatives pointing out the threat of having too much data but too little data quality and even less information. In a related vein, Gartner Inc recently reported that 85% of AI projects in companies fail to deliver on the promised return partly because of lack of high-quality data, the right metrics, actionable recommended changes, and failed knowledge transfer to ensure implementation (Medium October 9, 2024). Actionable analytics refers to the process of extracting insights from data that directly inform and drive decision-making, enabling measurable actions to improve performance or achieve specific goals. The four papers in this special session aim to contribute to making marketing analytics more actionable. Themes. The special session emphasizes the following four related themes: (1) key measures and metrics to overcome information overload and metric fatigue, (2) data and knowledge accumulation to enable generalization of results, (3) interactive data visualization to effectively communicate consumer and marketing insights, (4) machine learning integration to aide in integrating the first three themes. The papers in this special session explicitly address these themes to advancing actionable analytics. On the first theme, key measures and metrics: Paper 1 proposes new metrics to quantify mediation analyses. Paper 2 proposes improved metrics for meta-analyses. Papers 3 and 4 use and propose precise biometrics to identify attention patterns of consumers using eye-tracking. On the second theme, data and knowledge accumulation. Paper 2 proposes improved methodology for meta-analyses. Paper 1 builds on this when conducting meta-analyses on published mediation analyses. Paper 3 uses a unique, large meta-analytic database of eye-tracking tested ads. On the third theme, interactive data visualization. Paper 3 offers an interactive web-application to upload new ads-in-context, providing the (out-of-sample) predicted attention metrics and heat-maps of the predicted focal attention points in the ad. Paper 1 offers an interactive app that visualizes the distributions of mediated effects and enables interactive counterfactual analyses. Paper 4 returns visualized recommendations to adapting product packaging design to maximize their findability in cluttered shopping environments. On the fourth theme, machine learning integration: Papers 3 and 4 make steps forward here. Paper 3 integrates a theory of drivers of attention with several machine learning methodologies, extracts the key low-level (perceptual) features and high-level (conceptual) themes from advertisements to make robust out-of-sample predictions of attention by consumers to the ad and brand. Paper 4 advances the application of generative deep learning methodology and biometric data to address product design challenges in consumer markets. Jointly, improving on these four themes makes marketing and its analytics more effective and smarter too.