When all output steps have been completed, processing is finished at the document level. This final step is the backup step, otherwise known as post-processing. In this step, the backup worker performs final cleanup. One instance of this worker type is created for each tenant within the program. This instance is responsible for all scenarios assigned to the tenant. The backup worker accesses the same interfaces of the input systems as the input worker. The reason for this is that, in these input systems, the input data must be deleted or the processing status updated.

Analogous to the corresponding input worker, the backup worker works on the batch level: the backup worker cannot begin work for a batch until all documents in that batch have been completely and successfully processed. If an error occurs, the batches will remain in the system with an error status. The administrator will usually need to intervene, and processing must be attempted again.

All backup worker activities described below are options that can be enabled or disabled. Normally, the main task of the backup worker is to perform the actual backup of the input data to a file system folder or folder structure. If the original input of the data has already been performed via the file system, this original data will be used for the backup. Otherwise (e.g., in the case of a mail or web service input interface), the copies of the file attachments created in temporary storage will serve as the source for the backup.

After successful backup, the input data can be deleted if the input system supports this function. The data can be removed from temporary storage. Records are not deleted immediately from the administrative database at this point; they remain in the database for the purpose of providing information on successful completion of processing. Both records and old backup folders can be deleted once they reach a certain age. This job is performed by scheduler jobs, which must be configured separately.

If an error occurs, the backup worker sets error status in the input data. It performs this task even if the actual data backup is not yet to take place in this context.