Quantitative Analyses in Digital Marketing and Business Intelligence
This work is divided into two parts. The first part consists of four essays on questions in digital marketing; this term refers to all marketing activities on the Internet, regardless of whether they primarily address users of stationary devices (e.g., a desktop PC) or users of mobile devices (e.g.,...
Keine Tags, Fügen Sie den ersten Tag hinzu!
|Zusammenfassung:||This work is divided into two parts. The first part consists of four essays on questions in digital marketing; this term refers to all marketing activities on the Internet, regardless of whether they primarily address users of stationary devices (e.g., a desktop PC) or users of mobile devices (e.g., a smartphone). In Essay I, we model the time it takes until an item that is offered in the popular buy-it-now offer format is sold. Our model allows drawing inference from the observation of this time on how many consumers are interested in the item and on how much they value it. By this approach, several problems can be bypassed that often arise when these factors are estimated from data on items that are offered in an auction. We demonstrate the application of our model by an example. Essay II investigates which effects ads that are displayed on search engine results pages have on the click behavior and the purchase behavior of users. For this purpose, a model and a corresponding decision rule are developed and applied to a dataset that we have obtained in a field experiment. The results show that search engine advertising can be beneficial even for search queries for which the website of the advertising firm already ranks high among the regular, so-called organic search results, and even for users who already search with one of the firm’s brand names. In Essay III, we argue theoretically and show empirically that online product ratings by customers do not represent the rated product’s quality, as it has been assumed in previous studies, but rather the customers’ satisfaction with the product. Customer satisfaction does not only depend on product quality as observed after the purchase but also on the expectations the customers had of the product before the purchase. Essay IV investigates the relationship between the offline and the mobile content delivery channel. For this purpose, we study whether a publisher can retain existing subscribers to a print medium longer if he offers a mobile app through which a digital version of the print medium can be accessed. The application of our model to the case of a respected German daily newspaper confirms the existence of such an effect, which indicates a complementary relationship between the two content delivery channels. We analyze how this relationship affects the value of a customer to the publisher. The second part of this work consists of three essays that explore various approaches for simplifying the use of business intelligence (BI) systems. The necessity of such a simplification is emphasized by the fact that BI systems are nowadays employed for the analysis of more and more heterogeneous data than in the past, especially transactional data. This has also extended their audience, which now also includes inexperienced knowledge workers. Essay V analyzes by an experiment that we have conducted among knowledge workers from different firms how the presentation of data in a BI system affects how fast and how accurate the system users answer typical tasks. With regard to this, we compare the three currently most common data models: the multidimensional one, the relational one, and the flat one. The results show that it depends on the type of the task considered which of these data models supports users best. In Essay VI, a framework for the integration of an archiving component into a BI system is developed. Such a component can identify and automatically archive reports that have become irrelevant. This is in order to reduce the system users’ effort associated with searching for relevant reports. We show by a simulation study that the proposed approach of estimating the reports’ future relevance from the log files of the BI system’s search component (and other data) is suitable for this purpose. In Essay VII, we develop a reference algorithm for searching documents in a firm context (such as reports in a BI system). Our algorithm combines aspects of several search paradigms and can easily be adapted by firms to their specificities. We evaluate an instance of our algorithm by an experiment; the results show that it outperforms traditional algorithms with regard to several measures. The work begins with a synopsis that gives further details on the essays.|