A Comprehensive Framework for Assessing the Immediate Effects of Television Advertisements

Abstract

Precisely quantifying the direct influence of television commercials on online traffic is a complex endeavors. An exhaustive analysis is essential for a comprehensive understanding of consumer reactions to television advertisements. Nevertheless, this phenomenon has not undergone thorough investigation thus far. Prior research has employed either basic statistical tests or case studies to examine the responses of certain demographic groups, such as toddlers or a particular age range, to advertisements, or simple regression models to evaluate the effects of advertising. This paper presents TV-Impact, a comprehensive framework that utilizes the Bayesian structural time-series model called CausalImpact, together with unique supplementary approaches. TV-Impact incorporates a sophisticated algorithm that effectively identifies control variables, specifically targeting data sources that are not influenced by TV advertising. In addition, the idea of Group Ads was suggested to merge overlapping advertisements into a unified framework. In order to isolate the combined influence into discrete advertising affects, a Random Forest Regressor, which is a type of supervised learning technique, was utilized.

The TV-Impact framework was utilized to analyze data from iLab, a Turkish venture firm responsible for managing advertising strategy for its linked businesses. The findings indicated that the TV-Impact model had a favourable impact on organizations' allocation of TV advertising budgets and led to an increase in website traffic. It effectively functioned as a decision support system.