Creating a Data Factory for Data Products

Data is seen as the next big business opportunity. From a demand side, the popularity of artificial intelligence (AI) is growing and particularly deep learning requires large amounts of data. From a supply side, new technology, such as Internet of Things (IoT) sensors and 5G mobile communications, will greatly expand data generation. However, data has remained a challenge. In data analytics companies are struggling with too much time spent on data preparation. As of today, data preparation for analytics has largely remained handmade and made-to-order like cars before Henry Ford industrialized the auto business through productization of cars and parts, and factory automation. Similarly, for data analytics to become a bigger business, data has to be productized. First “data factories” are emerging to create such data products economically. This article introduces a framework to guide construction of a data factory: What are the key modules, why are they important, how is best practice evolving? The article is building on (a) a foundation and in-depth case studies in the literature, (b) current meta research and systematic literature reviews (SLRs), and (c) our own observations building a data factory. This real-world application uncovered the important additional steps of data rights management and data governance that may be less obvious from a computer science perspective but critically important from a business and information systems view.