Bid Estimation Methodology Based on Hourly Conversion Rate Forecast with a Classic Econometric Approach
Loading...
Date
2023-04
Authors
Suárez-Hernández, Laura F.
Journal Title
Journal ISSN
Volume Title
Publisher
ITESO
Abstract
Description
As emergence of auction-based digital marketplaces gains popularity, methodologies that help advertisers determine bid prices are receiving more attention. To better spend dollars is essential to understand consumer purchase dynamics and being able to estimate a bid that can leverage opportunity windows. The creation of a methodology to improve the bidding process based on a model at the keyword level which considers actual customer behavior through the day is needed. In this document we propose a mechanism to estimate the bid based on conversion rate forecast. We explore classic univariate econometric models using Box-Jenkins methodology and evaluate the performance comparing error metrics criteria. As result, we provide a comprehensive process and propose a methodology to calculate the suggested bid that displays dynamism and captures closer customer purchase behavior. Finally, we compared the current bid versus the bid suggested by our methodology to prove the difference on fluctuation and dynamism given both approaches. Actual versus predicted conversion rate are contrasted to corroborate it follows a similar pattern.
Keywords
Bid Estimation, Time Series, ARIMA, Digital Marketing
Citation
Suárez-Hernández, L. F. (2023). Bid Estimation Methodology Based on Hourly Conversion Rate Forecast with a Classic Econometric Approach. Trabajo de obtención de grado, Maestría en Ciencia de Datos. Tlaquepaque, Jalisco: ITESO