How it works
To replicate an upsert in Parabola for steps that do not natively support it:
First, pull in the data from the same location you are going to be sending data to. For example, if you want to upsert into a Google Sheet, start by pulling in that same Google Sheet.
Next, take the data that is ready to be upserted, and stack it on the data from the source step. Use a Stack tables step, set to stack columns based on their name, to stack these datasets. Be sure to connect the new data to the Stack tables step before you connect the matching source data. For example, when trying to upsert new data into Google Sheets, connect the new data to the Stack tables step before connecting the Pull from Google Sheets data to that step.
Next, use the Remove duplicates step to keep 1 instance of each row, based on some unique identifier column. By using this step after the Stack tables step, we are keeping all of the data from the new data, and adding in any old data that is not represented in the new data. At this point, we can push the data back out to something like Google Sheets, set to overwrite. Because both the new data and old data is present, the sheet will appear to just have the new data updated.Last modified on February 24, 2026
