Not the author, but if you have a write intensive workflow and I/O starts to be your blottleneck , compressing the data application-side allow your database to write less bytes, and thus to handle more load for the same capacity.

Look into ETL vs ELT

ETL you grab the data from somewhere, transform it to clean it/format it, then load it (in this case to postgres).

ELT, you grab the data, load it as is. Then you use another step to extract what you need or transform it. You keep the original data in case you need it there too.