Increasing complexity in the interdisciplinary development of smart products necessitates the integration of an increasing number of data sources for analysis scenarios like the development of digital twins. Ongoing trends like the digitization of engineering processes and smart products result in a large amount of data about products, their development process, and their usage phase. Using this data is not without its challenges. Especially in Small and Medium-sized Enterprises (SMEs), the data is typically stored in heterogeneous data formats and IT systems across several departments. Integrating data for specific application scenarios, like the development of digital twins, typically requires a large amount of manual effort. This contribution describes a process for the collection of data from heterogeneous sources across the product life cycle. The process is based on the phases of the Cross-industry standard process for data mining (CRISP-DM) known from data mining projects. It also describes a set of methods that support each step in the process of an envisioned scenario in SMEs with heterogeneous IT systems. In this context, the mapping of corresponding data objects across different sources is assisted by an Artificial Intelligence (AI-)based tool support. The method prescribes several steps to ensure the correctness of the AI’s suggestions and leaves the final decision to a human expert.