Would you say your advice equally applies to using Postgres with a JSONB column, which might make normalizing the data later on a bit easier (assuming it'd be k

Would you say your advice equally applies to using Postgres with a JSONB column, which might make normalizing the data later on a bit easier (assuming it'd be kept in Postgres)? Or is there something specific to MongoDB that would make it a better choice?