The optimization requires more information regarding the actual implementation of the python script that is being used to fetch the data and process it. Often times, the script slows down due to excessive use of loops, interruption and evaluation logic, and database calls done incrementally or procedurally. The optmization of the script in general terms usually requires prefetching and bulk fetching of database information pertinent to the merchant in question, then vectorizing the processing using arrays to minimize iterative processing of the data. Furthermore, pre-emptive indexing of the data using user_ids, timestamps, and transaction ids to minimize downtime during processing and data retrieval is also useful if not already done during the database call sequence, or module responsible for data retrieval. More in-depth solutions require advanced management of the python memory usage such as deduplication of data and objects being called and reduction of race based access in the data processing pipeline. Usage of sets and map based references reduce the amount of computational overhead, and subsequently, the time required to retrieve data for processing. If further performance is required, I also have experience in C++ and rust for even faster program requirements.