In the current process, the account statements are generated and exchanged between the customer and the exchange partner to match returnable packaging inventory.
Matching of account-statements has to be carried out by the receiver of an account-statement to check if all items in the account-statement match to the packaging-postings in the system of the receiver.
In the figure, the orange boxes depicts the activities for account statement creation, sending and receiving the statements for matching. The matching is depicted in green boxes.
The following features are provided in RPM for account statements:
- Generate Outbound Statement:
- manual PDF-generation by packaging-manager in app 'Manage Returnable packaging Accounts'
- Manual PDF-creation by partner in app 'Download statements'
- Send Outbound statements to exchange partners: automatic email-creation for all accounts of a partner with PDF-attachment via job 'Trigger Email Communication'
- Receive Inbound statements from exchange partners
- Creation via API, Excel-Upload or app 'Manage Inbound Statements'
- Matching
- Automatic matching with configurable weightages for each field
- Automatic matching based on machine-learning algorithm (GLIM)
- Manual matching by packaging-manager in app 'match inbound statements'
- Manual matching by partner in app 'review posting documents'
Steps in SAP Returnable Packaging Management
Automatic account matching is supported by scheduling a job, either for 'rule based matching' or 'GLIM based matching'. Open inbound statements are matched then according to the procedures defined in the jobs. SAP suggests NOT to run both 'rule based matching' and 'GLIM based matching' together and use either a daily, weekly or monthly period for the job.
The job is comparing the received inbound statements with the packaging-postings in the RPM system. Mappings from internal denomination to that of the partner can be maintained in the app 'assign exchange partner details' for business-partner IDs (including ID for plant, vendor and customer) and returnable packaging account, while material-number can be mapped in the app 'Manage returnable packaging accounts'.
Rule based matching requires to define a similarity-score as a percentage value from 0 -100% from which on a matching will be performed automatically.
This threshold-value and the weight that each field of an inbound statement adds to this similarity-score can be configured in the app 'configure weightages for matching'.
During the matching of each posting-item the similarity-score is calculated by adding up all weightages for fields where the value of the inbound-statement matches the value of the posting in RPM. Mostly an exact match is needed, but small differences in posting-date and quantity still add a partial weight to the similarity score. When matching the posting-date the lead-time is also taken into account.
GLIM based matching requires that a considerable amount of matched und unmatched inbound statements has already been processed in RPM. The GLIM algorithm is trained on this dataset to understand in which cases pairs are considered a match and when not. These self-learned 'rules' are then applied to the open inbound statements to perform the matching. No threshold values or weightages have to be defined in this case, but the model has to be defined via the app 'Manage Machine Learning Models' where it is possible to use the data from RPM as is (auto create model) or change the downloaded training data before using it for the training of the model (create manual model).
The result of the matching can be checked in the app 'Match Inbound Statements' where it is also possible to perform manual matching of inbound statements and see a comparison of the matched postings.