Shabnam Watson hits on a specific but interesting use case:
The most common scenario for setting up the out of the box incremental refresh in Power BI is to base it off of a datetime column; however, there are cases when you may want to set up incremental refresh based off of a column with a data type other than datetime. Examples are when you are working with a smart date ID (01012023 for Jan 1,2023) column or when you are working with a source system that has partitioned data using a column such as Year that has a numeric data type.
A use case for the latter scenario is when you are working with Parquet/Delta files via Azure Synapse Analytics Serverless SQL Pool. When working with larger datasets, it is typical to see the Parquet/Delta files partitioned by date ranges. Depending on how much data there is, the partitioning may be at the Year level instead of Day.
With that scenario in mind, read on to learn how you can minimize your Power BI processing time and costs when doing incremental refresh.