Abstract
Organisations invest considerable resources in advertising, with global spending on marketing in 2019 alone estimated at $618.7 billion (Handley 2019). In an increasingly competitive and noisy environment, accurately measuring the effectiveness of a company’s media channels and optimizing the resources invested in them has become a critical business problem. Businesses that can understand the effects of their advertising are well-positioned to improve the return on their marketing investments and gain a competitive advantage. However, developing and implementing data-driven solutions to quantify advertising media effectiveness is a challenging task. There is an increasing variety of data that can be collected about consumers’ interaction with media channels, such as display clicks, social media interactions and website page views. This leads to questions about how to best incorporate these consumer activity metrics, which are often used to represent owned and earned media channels, into models that quantify advertising media effectiveness. Additionally, many marketing managers struggle to interpret the results of statistical modelling and relate them to practical marketing calculations, such as media channel return on investment (ROI) and budget allocations.This research advances the field of marketing analytics by using consumer and firm activity data to develop statistical models that marketing managers can use to measure advertising media effectiveness. Using data provided by an Australian media agency, this research connects the outputs of these models with practical marketing calculations, such as media channel ROI and budget allocation recommendations. This provides marketing managers with a methodology they can use to better understand the effectiveness of their advertising and to optimize their advertising budget allocations. This research finds that disaggregating marketing performance metrics, as well as including consumer activity metrics and indirect effects in advertising media effectiveness models, reveals a more complete picture of the marketing environment. Including these relationships can enhance the accuracy of media channel ROI and budget allocation calculations.
This research also investigates how data used in advertising media effectiveness models can be applied in a different context to help address non-marketing business problems. In particular, this research examines how advertising data can be used to improve demand forecasts, which are important for strategic and operational planning. Producing more accurate demand forecasts can help companies become more profitable, efficient and effective. The results show that including advertising spend and calendar-based variables in time-series models can improve the accuracy of demand forecasts. Moreover, using relatively simple models augmented with exogenous variables can produce more accurate forecasts than more complex pure time-series models.
Since demand forecasts are often organised in a hierarchical structure, this research also examines how the performance of bottom-up and aggregate approaches to demand forecasting change across forecast horizons and as more data are collected. The findings suggest that when there is less data available, an aggregate forecasting approach produces more accurate forecasts, while a bottom-up approach becomes more accurate with more data. The results led to the development of an extension to time-series cross validation that reduces the sensitivity of results to the number of observations used in the initial training subset. This new technique, called repeated time-series cross validation (RTSCV), offers a more comprehensive way to assess time-series model performance.
Overall, this research leverages consumer and firm activity data to answer important business questions and improve strategic decision-making. The findings help researchers and industry professionals better interpret and translate advertising effectiveness models into practice. Tools are provided to help practitioners to better understand the effects of advertising and optimize marketing resources. Finally, this research demonstrates how marketing departments and media agencies can harness their data to provide additional value to clients by using it to address other business problems.
| Date of Award | 16 Jun 2021 |
|---|---|
| Original language | English |
| Supervisor | Bruce Vanstone (Supervisor) & Adrian Gepp (Supervisor) |