This appears to have worked! I first uploaded it as a csv but turns out data was lost so that didn't work. But then I saved it as a .xlsx file and uploaded that to databricks. I then made a table out of it and saw that all the attributes (contacts included) were of string type.

When running the above SQL command you posted, the output was returned with the contacts_parsed column and the little arrow allowing me to see the breakdown of the JSON object.

Then when running this as part of the larger query earlier with explode, I saw the emailIds column perfectly. My only remaining question is if this worked on contacts that had multiple emailIds. I don't know if there were any examples of this in the sample I gave you, but based on the solution you provided should it work on those contacts with multiple emailIds? "Working" would entail each row having one emailId, even if one contact has multiple emailIds, break them up so one is in each row of the emailIds column.