Auf der einen Seite wird geprüft, inwiefern die Notwendigkeit besteht, das Targeting als maßgeblichen Erfolgsfaktor im Dialogmarketing zielgrößenspezifisch auszurichten. Im Rahmen der vorliegenden Untersuchung werden zwei Bereiche hinsichtlich der Steuerung von Dialogmarketingkampagnen näher betrachtet. Given the chosen backpropagation optimizer, activation and cost functions, and considering the optimal number of hidden layers and nodes returned by the tuning process, the results showed not only that such non-linear structures are able to improve the quality of predictions, but also that the variables with the highest influence on the final purchasing decision were the ones related to customers’ personality treats and habits, to the detriment of the ones referred to their social network behaviour. For this purpose, several alternative networks – obtained setting different hyperparameters and inputs – were computed and evaluated in terms of out-of-sample performance. Multi-layer perceptrons are supposed to improve the quality of predictions with respect to traditional statistical models and to optimize the targeting activity avoiding addressing banners to non-interested customers. The aim of the research is to support marketing activities related to advertising banner-based campaigns on Instagram using feed-forward neural networks. Finally, results show that there are still plenty of interesting research possibilities, such as a comprehensive evaluation of models or new specifications of (mailing) variables. The majority of studies neglect latent heterogeneity and endogeneity. Furthermore, signs and significances of predictors vary across studies. Authors do not completely agree on which variables are the most important. As predictor effects are concerned results vary. However, Bayesian neural nets and Tobit models turn out to be good alternatives. Considering various studies of model evaluation it becomes evident that logit models frequently constitute a good choice.
Based on these modeling aspects we evaluate the different studies. Optimization methods are presented according to whether they refer to static or dynamic objectives. Besides, we analyze important modeling aspects, i.e., latent heterogeneity and endogeneity. Response models are divided into parametric and flexible models.
the latter distinguish static and dynamic effects. We classify various dependent and predictor variables and-w.r.t. Most of these studies analyze data sets from mail order companies or charities. Beginning with 1995, we discuss different studies that deal with response measurement and optimization of direct mailings.
0 kommentar(er)
