Similar to the software services methodology, data products (DP) can be seen as a kind of data services that specify all important issues for data provisioning and data consumption. DPs come in many different varieties stretching from simple data pipelines to complex machine learning models and model inferences and, above all, typically result in complex data networks. It is time to come up with a useful categorization and structuring of the DP topic in order to conquer complexity. In this paper, we present and assess a basic classification approach that focuses on DP characteristics and thus provides the basis for blueprinting and architectural discussions.