Compound join keys with null values

I am uploading a CSV that is assigning user events to different categories. In my data, the events have labels and sub labels. In some cases, the sub-labels don’t exist, so are null for that row. My categorization takes into account both the label and the sub-labels (when they exist).
I created a CSV with the following columns: labels, sub-labels, category.
I then joined to my product data using both label and sub-label as a join key.

The problem: only rows that have both the label and the sub-label are coming through. If the label matches, but the sub-label is null in both the data and the CSV, it isn’t bringing in the category.