In a full refresh, the extracted data is fully replaced with the data from the underlying data source, which can be time-consuming and usually runs slower. An incremental refresh, on the other hand, is designed to update the extract by adding only new rows of data, making it generally faster. Thus, the key differences between full and incremental refreshes are: a full refresh's replacement of all data versus the addition of only new rows, and the associated speed differences.