Finding creative drivers of advertising effectiveness with modern data analysis

The Modern Data Analysis paradigm (Williams, 2021) advocates using multiple methods to address the same research question, which is rarely done in studies of advertising creative effects. In this paper, we apply the MDA paradigm to data from Hartnett, Kennedy, et al. (2016), which coded 158 creative variables for 312 television advertisements with commercially validated short-term sales effectiveness outcomes. We found that many models give higher classification accuracy than the ordinal regression model previously applied, some significantly higher. Importantly, by applying many alternative but equally plausible analytical methods, we can identify creative variables associated with commercial success and have evidence-based confidence that these creative variables are artefacts of the data, and not artefacts of any particular analytical method and its associated assumptions. The findings reveal several alternative creative variables that are consistently associated with sales success across methods, which relate to the timing aspects of visual branding.