Incremental Refreshes - Unique Key Behavior with Delete+Insert Strategy?

[james-larsen]

Can someone help me better understand how dbt works with its unique keys when using a delete+insert strategy? My use case is to recalculate one record per customer per day for the next 7 days based on a number of factors. So technically, the unique key is customer ID and calendar date. However, during any given refresh, which customers are eligible to have a day calculated for them could change, and customers could drop off. If dbt works the way I hope it does, if I specify only calendar date as the unique key, I want it to determine the unique calendar dates in my incremental dataset, delete from the main dataset all matching dates, then insert the new customer / date records. This clearly would wreak havoc with a merge strategy, but I just want to confirm in this circumstance that providing only half the PK as a unique key isn't going to cause any weird behavior with the delete+insert strategy. Any insight would be appreciated!