Overcoming privacy issues by generating ‘fake’ customer data
Research Seminars Series | November 7th, 1-3 pm | TC 5.04
University of Groningen (NL)
Abstract: Privacy is a fundamental right of customers . Over the years, we observe growing attention for privacy concerns, due to wider availability of data and the the development of methodologies which jointly allow for more fine-grained and individual analyses of customers’ needs and wants. This leads to benefits for consumers, but this can also be perceived as pervasive. The associated privacy concerns have led to a critical attitude towards the collection and analysis of individual customer data. In this talk, I will discuss one of the possible directions to alleviate such privacy concerns. I will discuss a paper in which we develop generative networks that are able generate individual-level data of customers, that are non-existing, but mimic data of real customers. I will discuss three marketing data sets with different characteristics, to which we apply our approach. We show to what extent these non-existing, but mimic data of real customers. I will discuss three marketing data sets with different characteristics, to which we apply our approach. We show to what extent these networks are successful in generating ‘fake data’, by comparing the marketing insights they produce with the insights that the true data deliver. Surprisingly, we shows that analysis of fake data can in some cases outperform analyses of real data. We also pay attention to the possible opportunities that this approach can have for academics and firms to alleviate privacy concerns. We believe this might facilitate sharing data, even under tight GDPR requirements, thereby potentially improving business outcomes and accelerating scientific progress in the field of marketing.