What financial institutions and companies should know now

Until 2018, more than ten standards and formats existed in parallel in Switzerland's credit transfer and direct debit procedures. In addition, there were seven different variants of payment slips. Switzerland wanted to reduce this diversity. With the introduction of the European standards based on ISO 20022 in 2017, it reorganized the national payment traffic: The aim is to simplify the payment traffic procedures and the systems of the banks and PostFinance AG and also to align them with the European standards. Payment transactions should be more effective, faster and less prone to errors and also make money laundering more difficult. To this end, the previous proprietary account numbers were converted to IBAN. This also affected bank transfers, direct debits, payment slips and notifications.

Convert processes for payment processing
The new invoice with QR code simplifies payment transactions between billers and payers or their banks and the electronic data exchange in the interbank sector. The new invoices are intended to reduce processing times and costs and speed up processing by payment service providers and their exchange of information with each other. However, this can only succeed if all parties involved also adapt their IT hardware and software to the technical requirements. This means that they must upgrade their payment processing with QR code-enabled read and write technologies. The biggest

The challenge is to automate the processes in such a way that no manual processing is necessary. Here, numerous pitfalls become apparent in detail: In addition to the QR code, handwriting must also be recognized, contradictory information in the QR code and the text fields in the visual section must be clarified, and the interfaces to transaction partners must be adapted to the new possibilities.

Process several million QR invoices per year automatically
Several scenarios need to be thought through. In the simplest case of an invoice with the standardised Swiss QR code, there is a process chain in which a payer proceeds online: all the transaction data is on the invoice. On the one hand in the included QR code, but also in the text fields in the visible part. The payer scans the QR code with his banking app on his smartphone and triggers the transfer. There is no need for manual additions or changes to IBAN, reason for payment or invoice number and amount. By using the banking app, his personal bank details such as name and account number are already known. Since all the data arrives digitally at his bank, it can process the transaction automatically. The payee's bank receives the payment advice digitally and the payee receives the corresponding credit note. The order is debited from the debtor's account.

However, the payment process will often look different. Up to 20 percent of payment transactions are still paper-based and may be supplemented by handwritten information. And it is precisely here that financial institutions are still challenged to achieve a high level of automated processing. Especially financial institutions with several million payment transactions per year need powerful technologies for QR code processing (reading and writing), text recognition (including handwriting) and image processing as well as new processing algorithms.

In addition, there are new internal monitoring, verification and correction processes. If these processes are not properly automated, the Swiss QR code becomes a nightmare for the processors of a transaction.

Readout, validation and verification needs intelligent algorithms
For paper-based QR invoices, financial institutions need to extract the data they contain from the QR code and integrate it into their systems for further payment processing. For a smooth, automated process, the quality of the scanners is crucial. They must deliver sufficient image quality to transfer the data into their own data systems. Even if the paper is creased, dirty or yellowed, the algorithms must be able to recognize the information on the QR code completely and without errors within fractions of a second.

If the data are free of errors in the system, automatic validation and verification of IBAN, payee and payer must take place. Different spellings or transposed numbers must be corrected automatically, but the comparison with large databases must not lead to any delays. The allocation of discrepancies must be monitored; in case of doubt, the system must provide a warning so that employees of the financial institution can check the data manually.

Handwriting recognition as the biggest source of error
While reading and evaluating a QR code is rarely a problem today, handwritten information in the visible part of the invoice is the biggest source of error. Here, the accuracy and speed of the text recognition is crucial. While Optical Character Recognition (OCR) recognizes print letters printed according to the Swiss Code Standard unambiguously and without errors, handwritten names of payers and amounts require more powerful software for recognition. By means of Intelligent Character Recognition (ICR), further context information is included in the analysis and the results are corrected thanks to intelligent algorithms. For flowing handwriting, Intelligent Word Recognition (IWR) is also used.

Handwriting recognition as the biggest source of error
While reading and evaluating a QR code is rarely a problem today, handwritten information in the visible part of the invoice is the biggest source of error. Here, the accuracy and speed of the text recognition is crucial. While Optical Character Recognition (OCR) recognizes print letters printed according to the Swiss Code Standard unambiguously and without errors, handwritten names of payers and amounts require more powerful software for recognition. By means of Intelligent Character Recognition (ICR), further context information is included in the analysis and the results are corrected thanks to intelligent algorithms. For flowing handwriting, Intelligent Word Recognition (IWR) is also used.

Returns management with renewed QR code generation
If the payment order contains an inconsistency such as the incorrect number of payments or the sum of the individual payments does not match the total document, the entire payment order is returned to the customer. In such a case, however, the original receipts are not selected from the thousands that are processed daily, but the scanned receipts of the incorrect order are printed out and sent. After the invoices have been scanned by machine, the QR code is not of a quality that would allow the code to be read after printing. This means that the QR code must be reconstructed before printing. To do this, an identical code must be generated with a QR code generator, which overwrites the old QR code exactly in the right place. Image processing tools are used here to reduce deviations to a minimum. Only if this returns management is carried out meticulously can billers or payers correct their error and resubmit the supplemented or corrected payment order with QR invoice(s).

Billers must prepare ERP systems and accounting
Of course, it is not only financial institutions that are challenged by the introduction of the QR-bill, but above all the billers. Efficiency gains in processing can only be realized if the invoices are also created according to the specifications for the Swiss QR-bill. Ideally, the supplementary SWICO regulation is also complied with. This defines the contents in order to optimally support automatic invoice processing. The changeover is likely to be painful for smaller SMEs that currently work with a minimal infrastructure and ERP environment and make intensive use of the message field option. Today, this field allows any customer messages, even handwritten ones. This possibility of customer feedback will be eliminated without replacement. The new message field must be filled by the ERP and the content must be included in the QR code.

Conclusion: Only automation brings the necessary efficiency
The most advanced in adapting to QR billing are certainly the big banks and PostFinance, which have already prepared their systems for QR billing with internal and external expertise. However, all financial institutions must be able to process QR bills by 30 June 2020 at the latest. Those who have not introduced automated processes by then will have to reckon with massive additional expense in the medium term.

Companies that do not adapt their invoice processing to the new possibilities will not be able to benefit from the automation potential that will be possible in the future. Especially for those companies that already operate an application for the automatic extraction of invoice data, the new possibilities offer a major gain in efficiency and the opportunity for cost savings in the medium term.