SAP Leonardo and Customer Service Automation

Customer Service Centers are a huge expense and drain on businesses with volume service operations.

Using Artificial Intelligence and Machine Learning, Gartner estimates up to 85% of service calls will be automated by 2020

This is relevant to many industries: Government, Airlines, Banking, Insurance, Telecoms, Utilities, Consumer Electronics, and so on

Many areas of these businesses, like Marketing and Finance, can scale to manage ever-larger volumes of customers. However, the exceptions are the customer-facing operations, like a bank branch or call center, which requires a resourcing level in direct proportion to the number of customers being serviced. Therefore, these businesses often have the highest number of their employees engaged in customer-facing roles.

While service center employees usually attract only modest wages, by the sheer number of employees involved, the cost of running a service center can run from tens to hundreds of millions of dollars per year.

Therefore, even if Gartner and other pundits’ estimates are overstated, the ability to automate some services would still translate into a huge cost saving for customer service operations.

Alternatively, for those operations struggling to provide adequate services on a fixed budget, this offers scope to deliver a significant performance improvement.

Is this a problem? Do organizations care about alleviating the costs of service? Of course, they do! There have been many attempts to address costs in this area.

• Call centers have been moved from higher labor cost markets to lower labor cost markets, and with a high degree of success, but there have also been problems with geographical, language and cultural differences.
• Customer self-service applications have been introduced and these have been shown to reduce some service center traffic.
• Data wizards have been provided to help service agents through tedious provisioning. Expert systems have been introduced to assist agents in finding responses to customer queries.
• These are all approaches to reduce call handling times and improve quality of the customer experience. The problem, however, is that the human service agent is still the prime actor in providing customer service.

How can we automate service? What is the IT infrastructure required?

The prime actor that replaces the human agent in the automation story is the Chatbot. Chatbots are able to mimic human conversation using Natural Language Processing (NLP)

Chatbots can be accessed using Social Media like WhatsApp and Facebook Messenger. However, a Chatbot is only a kind of user interface. The ability to process a service request like a payment extension needs to be programmed. In this way, it mimics the role of the service agent by requesting information from the customer accessing the legacy application.

A Service Intelligence program is a machine learning application that uses past service transaction data to answer a service problem. The application can learn and improve over time. It is used to assist service agents, but when paired with a Chatbot, it can provide this capability directly to help end users.

Although powerful, there are limitations; not all service requirements can be anticipated, and the process usually allows for the chatbot to pass over to the human agent to complete the transaction.

Chatbots provide a new service channel and many businesses are introducing a chatbot into their customer service mix. The quality of these today is somewhat variable depending on the ambition and objectives of the provider. It would be fair to say that many of the Chatbots are limited in scope and capability and their main purpose is in trialing technology deployment and in customer adoption and experience.

Chatbots have certain advantages as a channel. They are superior to Apps in many ways as they do not have to be downloaded or updated. Navigation can be achieved without special knowledge, i.e., the user just asks a series of questions. In addition, the service can be conducted in the user’s timeframe and a record of the transaction can be maintained.

Notwithstanding the usefulness of chatbots, the prime channel in customer service is via phone and through a call service center. A chatbot can be enabled to manage speech by invoking Speech-to-Text and Text-to-Speech applications. In this way, the chatbot can talk to the user and provide a service response.

This final step paves the way for significant service automation. It is not anticipated that every call can be or should be managed by this scenario around chatbots. However, a lot of call center traffic could be alleviated with automation. This frees up human agents to deal with more complex problems.