Artificially intelligent helpers: chatbots in customer service

According to SEO experts, companies should use full sentences in their website texts and formulate headlines in the form of questions that potential customers could ask and type into a search engine. This tip is no coincidence: Intelligent search engines understand questions and complete sentences and deliver relevant results accordingly. As a rule, as soon as the first or second word is typed, the search engine recognizes what the user wants to know about it. Computers can now be programmed in such a way that they “anticipate” human intentions, decisions and the underlying structures, so to speak, and output tailored answers or data. And what's more: computers are able to learn. They remember previous searches, decisions, and patterns and can use them to optimize future results.

Bots, AI and humans: Key puzzle pieces in the contact center

This trend is reflected in chatbots, which currently appear to be omnipresent in the service sector. On websites of all kinds, we encounter small chat windows in which avatars offer us help with communication as assistants. Behind these windows are computer programs, the so-called chatbots. A chatbot is software that can conduct text-based dialogues with people. Using keywords, word combinations and synonyms, it performs rule-based if-then operations along a tree structure. This means that it provides answers to certain keywords or word combinations or content that has previously been stored in a library.

If the end of an answer tree is reached, i.e. if the chatbot no longer knows what to do, the software forwards the chat to a person. Or the AI-based software performs probability calculations based on predefined classifications, targeted queries to the chat user, or keywords and word combinations in order to identify the user's concerns and provide a suitable answer. Here, too, artificial intelligence makes use of a previously defined data pool. And she's learning too: Because the validity of her probability calculations increases with the number of calculations.

Chatbot, artificial intelligence and human agents — these three pieces of the puzzle in customer service are combined in any way from a technical point of view in the “Chat” service channel. Anyone who has ever used a chatbot knows that the technology works surprisingly well. It is usually hard for users to see whether their request is being handled by a person or a machine.

Customer service chat has long been mainstream

In particular, rule-based bots often provide answers to recurring questions or concerns that are easy to summarize with rules. In retail, these include questions such as “Where can I find shorts, blue pants for men? “, because the categories “short”, “blue”, “pants” and “man” can be easily read from the product catalog using if-then queries. A second example from the retail sector is the question of the status of an order. Here, too, the necessary categories (e.g. the customer number) can be linked relatively easily with a query from the inventory management system and the corresponding result delivered.

Insurers, in turn, can implement claims reports completely based on chatbots, for example. This is because all key data, including the insurance number, the date, the amount of costs and even images, can be recorded just as easily with the help of a text-based dialog as by telephone.

For companies, shifting simple customer inquiries from expensive communication channels such as the telephone to a cheaper channel (the bot) offers significant savings potential. In the case of insurance, for example, if you assume that an agent needs an average of ten minutes of working time to register a claim by telephone and that the price per minute for a request of this quality is around 1.50 euros, the costs amount to around 15 euros per contact. This does not yet include the opportunity costs for the customer's potentially frustrating wait time while on hold. The costs of working time required to record the claim in the system alone can be reduced many times over with the help of the bot.

Man and machine work hand in hand

Classification, in turn, is used for concerns whose direction cannot be precisely estimated, and in areas where the product range is constantly changing. Recognition of concerns with insurance companies is a typical case. The customer must first be assigned to a division of business, such as life, health or motor insurance. A comparable use case from the e-commerce sector would be the question asked by the bot “What are you looking for? ”. Here, the customer must first be assigned to a product group. For example, if he is looking for a new vacuum cleaner, it can be operated with a cable or battery, have a bag or not, be from a specific manufacturer, or the customer has specific price expectations.

In both cases, the bot uses confidence queries to find the appropriate issue or product. The so-called hybrid model, i.e. the combination of man and machine, is often used here. If the software is unable to unequivocally identify the issue, the chat will be handed over to a person. Depending on the requirements, the agent can then freely continue the chat or draw on ready-made answers.

Even with the more complex classification scheme, working time should be shifted to a more cost-effective communication channel and make customer communication more efficient. The software sometimes finds shorts for men much faster than a human agent.

Sometimes the number of ongoing business transactions can only be mastered with the support of a bot within a specific time frame (such as during peak hours). For example, thanks to a chatbot implemented by IP Dynamics, an insurer was able to fully process the peak of inquiries at the time of the insurance change within a short period of time. This would not have been possible only with the telephone channel. With the help of the chatbot, this customer generated sales in the six-digit euro range in commissions alone.

Chatbots are revolutionizing the contact center

The integration of chatbots as assistants not only offers business benefits, but also benefits consumers and thus has a positive effect on the company's image. The aim is to remain attractive for demanding, online-savvy and experienced target groups. Customers are looking for an easy way of communication that gives them quick answers around the clock. This is shown, among other things, from the “2018 State of Chatbots Report,” for which several companies surveyed a representative sample of the US population. According to this study, “24/7 service”, “get an answer quickly”, “get answers to simple questions” and “easy communication” are at the top of the list of the mentioned benefits of a chatbot from a user's point of view.

From a company perspective, customer care must become more cost-effective and efficient. Customers want to use simpler and faster communication channels in contact with companies and at the same time be better served. We are at the beginning of a major development that will transform customer service and the contact center in the long run. Contact centers that route to human agents may be completely replaced in the long run. Especially when it comes to artificial intelligence and machine learning as well as with regard to the use of emerging data for analysis, not all options have been exhausted yet.