Update async-cascading.md

Author: NHelgren

power-platform/admin/async-cascading.md

@@ -30,10 +30,10 @@ If an environment is encountering timeouts or degraded performance while the syn
 
 |Synchronous mode  |Asynchronous mode  |
 |---------|---------|
-|No other jobs can be executed on the entire set of selected records (direct or cascading) until the cascading operation is complete.   | Cascading changes are batched, locking only the records being processed within the batch. This allows other jobs to execute during the full cascading change operation.         |
+|No other jobs can be executed on the entire set of selected records (direct or cascading) until the cascading operation is complete.   | For Assign and Delete, cascading changes are batched, locking only the records being processed within the batch. This allows other jobs to execute during the full cascading change operation.  For Merge, the changes are still run as a single batch but done asynchronously to provide control back to the user more quickly.       |
 |When the job is completed, all data shows the new desired value.     | As the job runs, each completed batch displays the desired value. This means that there will be a time when some data shows the desired value and some shows the original value until the full operation is completed. This is referred to as “eventual consistency.”  |
 |If a single record fails, all data is rolled back to the original value. The rollback will require re-editing all completed records, which takes additional time.     |  If a single job fails, it is retried multiple times to attempt completion. If the job can't be completed the failure is recorded in the **System Jobs** area. Notice that successfully completed records retain the new value.       |
-|If one of the records in the cascading list has a value that is different than the expected value, the job will fail and roll back. For example, the starting record belongs to *Owner 1* and the cascading operation wants to change it to *Owner 2*. If one of the downstream related records has changed to *Owner 3* or is deleted before the lock occurs, the entire job will roll back.     | The operation always works in overwrite mode changing the current value to the new value based on the parent child relationship. There are no job failures due to an original value mismatch.        |
+|If one of the records in the cascading list has a value that is different than the expected value, the job will fail and roll back. For example, the starting record belongs to *Owner 1* and the cascading operation wants to change it to *Owner 2*. If one of the downstream related records has changed to *Owner 3* or is deleted before the lock occurs, the entire job will roll back.     | For Assign, the operation always works in overwrite mode changing the current value to the new value based on the parent child relationship, there are no job failures due to an original value mismatch.  For Delete and Merge, if a record that was expected as part of the set is missing, all the records up to the failure point are considered completed. The user or admin can re-execute the failed job which will recalculate the job to continue without the missing record.      |
 
 ## Asynchronous mode and plug-ins
 When a cascading transaction has more than 100 records and does not have any plug-ins associated with the records, the records will be processed asynchronously.