Not that long ago, customers service centers were only accessible via telephone and e-mail. Nowadays, customers reach out to organizations by a variety of mediums, for example WhatsApp and Social Media. Not only do customers choose their preferred medium of communication, they also expect and demand to get assistance instantly. In response to these high expectations, more and more organizations react by adding new mediums to their customer service centers and increase contact hours to enable on-demand services.
These developments have intensified the complexity of organizing the customer service center. Alignment between all available mediums is asking more and more from an organizational perspective. Organizations therefore need to rethink their customer service center to meet the expectations of their customers in an efficient way. In this article we will elaborate on how organizations can gain in efficiency by deploying Artificial Intelligence and Machine Learning. More specifically, how bots and human agents can work together to get the work done as efficient and as personal as possible.
The moments wherein organizations are in direct contact with their customers are limited in today's digital world, and often evolve around negative experiences such as complaints, payments and returns in retail and e-commerce. Not even speaking about the insurance sector. Contact moments with your insurer often take place after experiencing a negative event such as damage, loss and health issues. In these scarce moments of contact, it is key to exceed expectations. Simplicity, understanding, accuracy and speed are at the foundation of customer satisfaction. To create real valuable contact moments, organizations should deploy some level of intelligent assistance. The integration of assistance from AI is not just a mere luxury, it is inevitable. Scaling customer service in terms of human capacity is not the solution, eventually expenses will become untenable and economies of scale are very limited. These days, we see a rise of companies exploring and implementing the possibilities of intelligent assistance in their customer service center successfully. Gartner predicts that 25% of customer service operations will use virtual customer assistance by 2020 (Gartner Customer Experience Summit, 2018).
Negative events in insurance
The helping hand of Artificial Intelligence can be deployed in various ways. In this article we will focus on ‘suggested answers’, ‘question prediction’ and ‘intelligent capacity planning’.
Before we dive deeper in the topics outlined above, we first need to discuss the traditional chatbot. Customers are in contact with chatbots more then ever. These bots are providing answers automatically, resulting in instant responses and reducing the workload of human agents. The benefits are clear, but the same holds for the downsides. These chatbots often work based on triggers and/or cues. In other words, if a certain word is written a predefined answer is provided. You can imagine that the chatbot isn't able to have any sense of context and therefore is not always providing the right answers or nuances. Another aspect that the Chabot is lacking, is a personal touch, the understanding of a fellow human being. Customers still prefer human interaction for more complex and personal/emotional questions and issues, obviously a chatbot can’t live up to these expectations. Lastly, a chatbot isn’t scalable in the long run. The triggers needed for the working of a chatbot need to be set by IT employees. This is a very tedious and time-consuming task and it needs to be updated regularly. In conclusion, an independent self-regulating chatbot has to many shortcomings to fully trust on.
So, how can we leverage the power of Artificial Intelligence to efficiently handle customer interactions while at the same time mind the needs of the customer. The answer can be found in intelligent assistance. With intelligent assistance the human agent is in control and makes the final call. It’s the collaboration between human and machine that will significantly increase efficiency while at the same maintain the personal touch.
In practice this means that when there is an incoming question, the smart algorithm is analyzing the text in real-time and can assist the human agent by giving two types of proposed answers: a suggested answer based on the collective memory of all historical questions and the corresponding answers given by human agents in the past. The algorithm predicts which question was most similar in the past and provides the answer that was given by the human agent at that time. Or, the algorithm can provide a ‘standardized’ answer on various clusters of questions by predicting the topic of the question like the traditional chatbot explained earlier. However, in this case the bot only suggests these two answers and don't put them directly into the conversation. The human agent can decide if the answer is suitable to the question or not, and if necessary adjust the answer to the context or provide a entirely new answer. See the videon below for a better understanding of this collaboration.
In this way the AI does not take over the conversation, but is providing the agent with relevant input to enhance the speed and accuracy of the conversation. Because the human does not have to think about or search for the correct answer, but only has to place it in context and add human emotions.
By accepting and or adjusting the proposed answers, the agent is in fact giving feedback to the bot. By capturing this feedback in the collective memory and using it in future predictions, the algorithm will become smarter over-time and will provide more accurate answers in various contexts.
The learning algorithm becomes smarter over time
Based on the collective memory of trends and patterns in coversations, the next question of the customer can be predicted before the question is asked. In this way, the human agent gets the opportunity to proactively help and respond to customer needs. This really gives the opportunity to exceed customer expectations.
Intelligent assistance will decrease the number of interactions and time spent per interaction, while at the same time increase customer satisfaction by focusing on personal interaction and connection.
Furthermore, the insights gained by intelligent chatbots can be used to offer the customer a better service through other channels. For example, if certain questions are asked frequently in a given timeframe, they can be added to the FAQ or in mailings. In this way a dynamic FAQ can be created that is always aligned with the real-time concerns of your customers. It enables companies to detect problems, for example with a product, in an early stage.
Intelligent customer service is not only beneficial for individual interactions, it also provides insights on an aggregated level. Based on trends and patterns that are distilled from the incoming data, topics and subjects that are going to be asked by customers in the near future can be predicted. Capacity planning can hugely benefit from these accurate forecasts. For example, if peaks in demand are predicted, capacity can be adjusted accordingly. By doing this the capacity of human agents can be optimally aligned with demand. This will prevent over capacity in times of low demand and guarantees service quality during peaks. Based on text analysis the topic of incoming questions can be determined and categorized in a way that customer requests can automatically be routed to the right employee in the contact center. The customer always gets in contact with the right person with the right knowledge immediately. Besides, human agents can be better instructed to provide the right answers.
The above proposed solutions show that the crux of the matter is not the use of a single solution, but the combination of both. By doing this, personal interactions are improved by more accurate and faster answers and customer service will become more efficiently on an organizational level.