Many see the capability of chatbots to engage in dialog as their biggest potential in customer service applications. To this day, however, smooth communication with a machine via free text without loss of context is not possible. Nevertheless, bots already now provide added value for customers if their existing dialog skills are combined with technical processes carried out autonomously by the bots to solve customers’ issues. Telecommunication companies can benefit from this option to generate a positive customer experience outside of call center business hours.
Comparing the requirements made in a bot (understanding customers’ issues, pre-qualifying, finding approaches to a solution, and rectifying the problem via dialog) with what has actually been implemented, it becomes evident that optimization is necessary: According to a survey, 27 per cent of the participants said that bots were not intelligent enough to answer questions efficiently, and 24 per cent thought that the bots do not take sufficient note of the context. 47% would abort a session if the bot cannot answer their question. So, no top grades for the bots’ language skills. But the study also comes to the conclusion that robots can provide fast and convenient help to customers under certain circumstances. Companies can use this potential by having the bots execute technical processes: In this case, successful solving of a customer’s issue does not depend on the machine’s language skills alone. In particular telecommunication companies with their huge volume of technical processes can improve their customer service in this way.
A bot is process capable in that it tests technical features, such as the functionality of a router, and rectifies the problem if applicable. This additional function makes the chatbot valuable for customers since it can speed up and optimize technical processes outside of the usual business hours. Fast reaction and solution times are a key factor to customer satisfaction. However, the positive outcome of a bot application depends on preparatory work: the clear definition of the bot’s application field and of the use cases which should be covered.
Language is the sine-qua-non. This condition is met by use of a commercially available NLU (Natural Language Understanding). Available via the Internet, these language systems allow the bot to understand language. Pre-fabricated reply buttons, which are frequently seen with Facebook bots, help the machine to communicate with the customers because there is less free text to interpret.
Language is the sine-qua-non. This condition is met by use of a commercially available NLU (Natural Language Understanding). Available via the Internet, these language systems allow the bot to understand language. Pre-fabricated reply buttons, which are frequently seen with Facebook bots, help the machine to communicate with the customers because there is less free text to interpret.
The bot needs knowledge to pre-qualify and define the cause of an issue. This knowledge should be available centrally in a kind of knowledge source or base. This has several advantages:
Experienced service technicians supply content to a well-structured knowledge base. A sufficient volume of high-quality data on previous customer issues and their solution will make it truly efficient: Now, even better content for solving of customer issues can be generated by means of machine learning (ML) and automation.
To solve a customer’s issue, the chatbot needs a solution procedure assigned to a specific combination of issue and cause. This procedure should be stored in the form of processes or scripts. If this condition is met, the chatbot not only is capable of engaging in dialog but also of executing technical processes. Thus, the chatbot can autonomously complete all the steps from understanding a customer issue to solving it by means of dialog and process execution. If all these technical preconditions are met, the chatbot will provide true added value for the customers.
- A chatbot provides true added value if it is not only capable of engaging in dialog but also of executing technical processes for solving customer issues.
- Certain technical requirements must be met to allow the bot to do this, for example a central knowledge base with stored scripts and processes. For a more efficient bot, the knowledge base should be fed with content generated by ML and automation.