Advertising exerts an immediate influence on sales, as well as delayed effects that persist over time. The infinite geometric distributed lag, commonly referred to as the KOYCK model (the ARMAX alike representation), is deployed extensively (Franses and van Oest, 2007).. The conventional method for estimating model parameters involves implementing a grid search and recursively constructing the infinite stock series (Berkowitz et al., 2001). The alternative representation of the KOYCK model naturally introduces a different estimation technique, which is considered more robust in accurately retrieving the true parameters. This alternative representation also makes it possible to implement a robust dynamic extension, which is something that is utilized in these applications much less frequently. Variables in online search engines serve as proxies for the responses of intermediate users. This response is influenced by advertising and, in turn, drives sales. The underlying mechanism results in a dynamic simultaneous system that facilitates the decomposition of the expected advertising returns between sales and the intermediate response. This thesis focuses on developing the innovative estimation technique for practical applications, broadening the model to encompass various advertising channels and applying the estimation method in this multi-dimensional context. The simultaneous system is presented as a method for comprehending the dynamic influence that advertising exerts on natural search engine variables. The developed models are subsequently employed to examine the returns of the different channels, aiming to illustrate the application of these less conventional models.
La pubblicità esercita un'influenza immediata sulle vendite, così come effetti ritardati che persistono nel tempo. Il ritardo distribuito geometrico infinito, comunemente indicato come modello KOYCK, è ampiamente utilizzato (Franses and van Oest, 2007). Il metodo convenzionale per stimare i parametri del modello prevede l'implementazione di una ricerca a griglia e la costruzione ricorsiva della serie infinita di azioni i (Berkowitz et al., 2001). La rappresentazione alternativa del modello KOYCK introduce naturalmente una diversa tecnica di stima, che è considerata più robusta nel recuperare accuratamente i parametri reali. Questa rappresentazione alternativa consente anche di implementare un'estensione dinamica robusta, che viene utilizzata in queste applicazioni molto meno frequentemente. Le variabili nei motori di ricerca online fungono da proxy per le risposte degli utenti intermedi. Questa risposta è influenzata dalla pubblicità e, a sua volta, guida le vendite. Il meccanismo sottostante si traduce in un sistema dinamico simultaneo che facilita la scomposizione dei ritorni pubblicitari attesi tra le vendite e la risposta intermedia. Questa tesi si concentra sullo sviluppo dell'innovativa tecnica di stima per applicazioni pratiche, ampliando il modello per comprendere vari canali pubblicitari e applicando il metodo di stima in questo contesto multidimensionale. Il sistema simultaneo si presenta come un metodo per comprendere l'influenza dinamica che la pubblicità esercita sulle variabili naturali dei motori di ricerca. I modelli sviluppati vengono successivamente utilizzati per esaminare i rendimenti dei diversi canali, con l'obiettivo di illustrare l'applicazione di questi modelli meno convenzionali.
Modeling advertising effectiveness
Ameen Hayder Mohamed Elzaki
2024/2025
Abstract
Advertising exerts an immediate influence on sales, as well as delayed effects that persist over time. The infinite geometric distributed lag, commonly referred to as the KOYCK model (the ARMAX alike representation), is deployed extensively (Franses and van Oest, 2007).. The conventional method for estimating model parameters involves implementing a grid search and recursively constructing the infinite stock series (Berkowitz et al., 2001). The alternative representation of the KOYCK model naturally introduces a different estimation technique, which is considered more robust in accurately retrieving the true parameters. This alternative representation also makes it possible to implement a robust dynamic extension, which is something that is utilized in these applications much less frequently. Variables in online search engines serve as proxies for the responses of intermediate users. This response is influenced by advertising and, in turn, drives sales. The underlying mechanism results in a dynamic simultaneous system that facilitates the decomposition of the expected advertising returns between sales and the intermediate response. This thesis focuses on developing the innovative estimation technique for practical applications, broadening the model to encompass various advertising channels and applying the estimation method in this multi-dimensional context. The simultaneous system is presented as a method for comprehending the dynamic influence that advertising exerts on natural search engine variables. The developed models are subsequently employed to examine the returns of the different channels, aiming to illustrate the application of these less conventional models.File | Dimensione | Formato | |
---|---|---|---|
Thesis_Ameen Elzaki.pdf
accessibile in internet per tutti
Dimensione
2.01 MB
Formato
Adobe PDF
|
2.01 MB | Adobe PDF | Visualizza/Apri |
Executive_Summary_Ameen Elzaki.pdf
accessibile in internet per tutti
Dimensione
559.89 kB
Formato
Adobe PDF
|
559.89 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/235178