Data-driven innovators – An empirical analysis of data-driven SMEs and start-ups

Abstract

In today's fast-paced technological landscape, the adoption of Big Data Analytics (BDA) goes beyond incremental improvements in productivity and efficiency, enabling the creation of new products, services, and even business models, known as Data-Driven Innovations (DDI). As technology evolves rapidly, reshaping industries and our daily lives, businesses must adapt to survive and innovate to thrive. Entrepreneurs, in particular, have a vast horizon of possibilities to explore through data, opening avenues for new ventures. However, the literature in the field has pointed to a lack of empirical evidence on the actual realization of these possibilities, the so-called ‘deployment gap’. Also, regarding established firms, there is a recurrent call for longitudinal analysis to understand the dynamics of BDA adoption and firm performance. Motivated by these challenges, this study employs a data-driven methodology that integrates data science techniques, like web scraping, natural language processing (NLP), and neural topic modeling (BERTopic), to provide large-scale empirical evidence on the realization of the DDI, focusing on understanding the firms behind them. The objectives range from identifying data-driven firms using website text to analyzing the determinants of adoption, firm performance dynamics, and emerging business models in startups. The study starts by focusing on German knowledge-intensive SMEs and identifying factors influencing BDA adoption, following the Technology-Organization-Environment (TOE) Framework. The findings show that larger, younger firms with international ownership are more likely to adopt BDA technologies, and this adoption is positively associated with innovation indicators such as patents and trademarks. The second study, grounded on Resource-Based-View (RBV), extends the analysis by exploring the timing of DDI deployment and its impact on firm performance over time using panel data. The results show that early adoption confers performance gains, particularly in technology-intensive sectors, but these gains tend to decrease as the technology becomes more widespread. The third study shifts the focus to the global start-up ecosystem, analyzing emerging data-driven business models (DDBMs) by examining the value propositions of start-ups across various sectors. Using neural topic modeling, the research identifies key trends and patterns in DDBMs, confirming the increasing emphasis on AI and data science as central themes. The study also tracks the evolution of these trends over time, identifying a shift towards more specialized technological areas within start-ups' value propositions. The empirical findings contribute to the broader discussion on BDA technologies, innovation, and their influence on firm performance. They offer insights not only to researchers conducting qualitative and theoretical studies but also to practitioners and policymakers involved in technology adoption and entrepreneurship. Methodologically, this work contributes to innovation studies by applying advanced data science techniques to analyze large-scale, unstructured data. These methods introduce a novel approach to uncovering patterns and insights that traditional methods may overlook, thereby advancing the study of digital innovations.