I have survey data that is stored in JSON format. There are multiple versions of the same survey. One may have more questions then the other. First i do a json parse to get name/value pair columns alone with the survey name and version number. I then use a crosstab to pivot the data and finally a union to combine the multiple versions. My issue is all of the result columns (questions) are in alphabetic order. I want to maintain close to the original order.