Sure, but if the data is already there, it's a sifting and pruning problem, which can be done after ingestion, if needed.

It's better to have all data and not need it, than to need it and not have it. Assuming you have the resources to ingest it in the first place, which seems like the focus of the optimization work they did.