Unfortunately that's not how incremental refresh works. It assumes the query is constant, but the data in the database changes. So when you change the query, it starts over.
Is there a reason you aren't just extracting it all in one go? 500 million records is a lot, but not impossible.
If you really want to do what you are trying, see Add Data to Extracts - Tableau
Thanks for that clarification Dan!
The reason is purely limitations on how long queries can run. The admins have processes in place that kill queries that run X seconds.
I was hoping to avoid going through all the tape on these limitations by finding a workaround.
I looked at the link you sent. I had not thought of appending in such ways to my extract. If in my shoes is there a way you'd recommend attempting that approach?
Think I am slightly confused on going about slowly appending data but keeping my Custom SQL in the format desired for the final iteration. If using Custom Sql at all is feasible using that approach.
I'd try it in one pass. If that doesn't work because the admin kills the query, then see the article I sent. You can download each month separately and append them on the client.