Sale effects of attention to feature advertisements: A Bayesian mediation analysis

There is much evidence that the presence of a feature advertisement can increase the sales and market share of the featured product. However, little is known about how feature ad characteristics (e.g., size, color, and location of the advertisement) affect the sales outcomes and how the effects take place. Prior research has predicted that feature advertisements lead to behavioral outcomes through their effect on consumers' attention. Building on this idea, the authors propose a Bayesian statistical model to study how feature ad characteristics affect sales of the featured products and the mediating role of attention in these relationships. They use data from eye-tracking tests of feature advertisements, aggregated and matched with sales data at the level of the feature advertisement. Their approach accounts for endogeneity in the key variables involved and overcomes limitations of standard mediation analyses. They show that the gaze duration on a feature advertisement affects sales of the featured product beyond the mere presence of the advertisement and that a standard mediation analysis that does not accommodate endogeneity produces biased estimates of the effects of feature ad characteristics on sales. Their proposed methodology is widely applicable to mediation analyses. The findings imply that attention data collected in lab tests can help marketers compare the relative sales outcomes of different feature ad designs and improve the effectiveness and efficiency of feature adverting decisions.