Efficient and accurate processing tracking of customer enquiries is crucial in terms of customer satisfaction and business success. Answering these enquiries is often done manually and is therefore linked with high personnel costs in the sales department. In a first step, this paper analyses the customer enquiries of a representative component and system manufacturer. In a second step, the paper compares four different concepts for the automatic extraction of relevant data based on real customer enquiries to generate a product recommendation. A hybrid approach consisting of a lexicon, regular expressions and a Large Language Model delivers the best overall results.