Every successful run of a job creates a new restore point that lets you return your data to an earlier point in time. When you define retention policy, you specify how many restore points you want to keep and thus how ‘far’ you want to be able to roll back. Once the specified number is exceeded, the earliest restore point will be automatically removed. So if the retention policy is set to three and you already have three restore points, the fourth successful run of a job will delete the restore point created at the first job run