Digital Analytics: Linking Online Data to Purchase Behavior and Firm-, Stock- and Political Performance
(A2019-8708)

AUTHORS

Evert de Haan, University of Groningen

ABSTRACT

In this session, four studies that focus on getting insights from digital data, from click-stream data to social media and online search data, are brought together. These data are used to explain online purchases, as well as and firm-, stock- and even political performance. The studies provide an overview of novel ways and techniques to get insights from these data in order to explain marketing phenomenon and relevant outcome variables for firms and society at large.

1. Vices Go Together: Product Purchase After Online Compulsion

Radek Karpienko (WU Vienna), Anatoli Colicev (Bocconi University), Thomas Reutterer (WU Vienna), Koen Pauwels (Northeastern University)

Digital addiction lacks clear connections to marketing theory and is supposed to have dramatic consequences on purchase behavior. Using clickstream data, we model purchase behavior after consumption of “vices” on a browsing session level. Our modeling results show that compulsive users generally exhibit a higher purchase probability than households with medium or low compulsion scores

2. Search Engine Visibility and Shareholder Value: Financial Markets Valuation of a Firm’s Visibility in Organic Search

Gabriela Alves Werb (Goethe University Frankfurt)

This study provides a theory-based conceptual framework to explain the link between a firm’s visibility in organic search and shareholder value. Using a panel of publicly listed U.S. firms and organic search visibility, the author finds that a 1% improvement in a firm’s search visibility translates into a .01%-point increase in shareholder value.

3. Fake News and Real Events: What Drove the 2016 US Election?

Koen Pauwels (Northeastern University), Ginger L. Pennington (Northwestern University), Raoul Kubler (University of Münster)

We study the 2016 US election by investing potential driving factors by linking social media data, news articles, sentiment-, topic- and linguistic analyses, and major events to probability poll- and consensus polls data. We find that hate news, not fake news, drove Clinton’s polls and the effects of the candidates language on social media has often opposite effects on the probability poll versus the consensus polls.

4. Satisfaction Surveys or Online Sentiment: Which One Best Predicts Firm Performance?

Evert de Haan (Goethe University Frankfurt)

This study investigates if sentiment of online word-of-mouth is a good alternative for survey based satisfaction data in predicting firm performance. Although the initial results indicate that traditional survey data yield better predictions, online sentiment does add incremental value. Combining survey data with sentiment data can improve predictions and help create a better dashboard of the customer base.