Smart customer service

Every month, the service department at Walbusch processes tens of thousands of customer enquiries via a wide variety of input channels – over the phone, via digital channels (such as e-mail or the contact form on the website), and offline by taking down a customer enquiry in store.

Processing the various enquiries involves a great deal of manual effort. This results in long waiting times during periods of high request volume, which, in turn, leaves a negative impression on the customer.

Customers most frequently requested to unsubscribe from receiving advertisements by post. In terms of digital delivery channels, this enquiry appears simple: While unsubscribing to an e-mail newsletter can be done by the customer at the click of a button and automatically stored in the system, it is not that simple for postal mailings.

The customer would have to make a call, write an e-mail to customer service, or use the contact form on the website. As texts entered individually by the customer cannot be evaluated reliably for these channels using rigid rules, a rule system to filter and recognize enquiries was not a feasible solution.

BSI compiled a solution approach that focuses on processing incoming e-mail enquiries for advertising reduction.

The technical implementation was based on an automatic request recognition, which operates on text recognition using an artificial neural network. The objective: achieving a significant reduction in processing time and costs by means of automated processing.

Once an enquiry for advertising reduction has been recognized, the BSI Customer Suite activates the advertising block in the correct channel for the customer in question. The customer also receives a message that their request to block advertising has been processed and the sending of postal material has been stopped. For greater transparency, the message contains a note stating that this enquiry has been processed automatically. If the advertising block has been recognized wrongly and does not correspond to the customer’s request, the customer can deactivate the block at any time with a simple response. In addition, the AI used can be continually trained further with new data and thus improved.

The customer inquiries processed by the service department in recent years and stored in BSI were used for the necessary training of the AI. As a result, the AI is optimized for the wording used by Walbusch customers and can identify their concerns in the best possible way. It is remarkable that the AI was able to recognize requests to unsubscribe from mailings with a hit rate of over 90 percent after just a few hours of training.

Reducing the workload for staff by automizing service processes means that they can use the time thus gained to provide personal support to customers – a key aspect of customer communication for Walbusch. Since the implementation of the project, the AI has successfully processed over 40,000 e-mails within six months, which makes it much faster and more efficient than the manual processing of these enquiries by a human could ever be.

Using the BSI Customer Suite pays off in several ways.

The increased automation is reflected in considerable time and cost savings. In addition to using well-known models such as ChatGPT from OpenAI, BSI’s solution stands apart compared to competitors by the fact that BSI can integrate existing open-source models for text and request recognition in a configurable manner and with very little effort into existing processes and execute them in the customer’s instance.

This means a high level of transparency about where data is processed and guarantees that legal data protection requirements are met. Moreover, an open-source model guarantees an insight into which data is used to train the neural network and ensures that data processing is in accordance with ethical requirements.

For Walbusch customers, the new process ensures an all-round positive customer experience thanks to the immediate processing of and automatic response to their enquiry. The investment is also paying off financially for Walbusch: After less than a year in use, the project achieved a positive return on investment.

Based on the positive experiences with the AI implementation, further service processes were optimized with the possibilities of the AI integrated in the BSI Customer Suite.

In addition to requests for advertising reduction, the area of application of AI now includes the prioritization of sales-relevant enquiries, the processing of which has a high priority at Walbusch due to their cost-relevance. For instance, changes or cancelations of orders as well as address changes are automatically filtered out of the incoming communications and assigned to the relevant service staff. This saves costs by avoiding unnecessary mailing or returns, is more sustainable, and makes the customer happy.

As a result, the incoming enquiries processed, sorted or prioritized by AI are now around 25% – and counting.