The potential of data mining and machine learning in the development of new products is acknowledged in both research and industry. Despite the increasing amount of data available in companies and the growing complexity of data and information flows, which are stretching manual and ru-le-based approaches to their limits, data-driven methods are still hardly used in product development. In this article, the un-derlying causes are examined in more detail and the results of an industry survey on the current challenges of data-driven methods in product development are presented.