Comment by CharlesW

Yes, I feel like this would've been more helpful generalized as "Consider DSV" (delimiter-separated values) than CSV specifically, because of the interop issues that often come up. I'd have also mentioned using Parquet.