Implementing a chatbot in Customer Service may be your best or worst decision. It all depends on the extent to which the strategy you choose will respond to your customers’ expectations. The market is clearly dominated by two fundamentally different approaches.
How to choose the right one to increase the satisfaction of your customers and employees?
IBM claims that 80% of routine cases reported to Customer Service can be handled by a bot. In other words, implementation of a bot can reduce maintenance costs of the department by up to 30%!
Globe Telecom, one of the largest telecommunications service providers in Southeast Asia, after the implementation of a chatbot, recorded a 22% increase in customer satisfaction, half the number of reported hotline calls and a 3.5-fold increase in employee productivity.
Such results are possible to obtain by implementing a bot correctly, and are probably achievable in your Customer Service department as well. So how not to make a mistake when choosing the solution?
Two bot implementation strategies
Currently, there are two types of chatbot implementation strategies in Customer Service departments: rule based and AI based. The choice between them should depend on the specificity of your industry and the nature of the inquiries that reach the customer service department.
Rule based chatbot
Rule based chatbot is a type of bot in which the dialogue with the user is based on successive choices from the available options. This means that the client does not type in the chat box as when messaging a consultant, but selects the next answer from the displayed options. The concept resembles a conversation, although most often the user is not able to enter his or her own questions. Such bots are most common on the Facebook Messenger platform.
One of the advantages of such a solution is its high simplicity. The process of implementation, depending on the complexity of the bot, takes no more than two weeks. The cost incurred is not high, including the payment for an appropriate platform for building the bot. Integration with internal CRM systems is also easy due to open API.
A consequence of such a simple implementation is the chatbot’s low level of complexity. Rule based bots only automate simple tasks that customers can often do themselves, e.g. handling returns or checking the status of a shipment. They can also answer simple questions about, for example, opening hours or directions. Statistics show that rule based chatbots help to automate about 30% of queries directed to Customer Service.
It is the more common solution, and the customers have already got used to it.
In the case of an AI based chatbot, interaction with the user consists in a dialogue. The user can enter their own queries, and a bot with a natural language processing engine analyzes their request and responds with a previously prepared answer. There is no problem in understanding the context of the sentence, so that the client has the impression of having a conversation with a real consultant. If the bot does not understand the request, it will notify a human consultant who can join the conversation at any time and respond quickly to the customer.
This solution is characterized by a higher level of query coverage. According to statistics, it reaches even 80% or maybe more, because the bot is continuously developed. Every conversation in which a consultant replies to a question previously unknown to the bot is saved and used to increase the bot’s knowledge. This happens during the consultant’s daily work and increases the effectiveness of the bot over time.
The implementation is obviously more complicated and the whole process, along with building the bot’s basic knowledge can take more than a month to complete. On the market there are platforms for creating conversation bots (e.g. DialogFlow), although usually this type of implementations is dealt with by companies that have their own solutions and offer comprehensive implementations.
What to consider?
As you can see, an AI based chatbot is a more complex and comprehensive solution. It has more to offer, but the costs of its implementation are higher. Often, however, the functions of such a solution do not determine the customers’ satisfaction. So how to choose the right option?
Nature of requests
The first step in choosing the right implementation strategy is to determine the nature of customer service requests. If the cases are complicated and consultants’ help is based mainly on conversation and e.g. searching for the cause of the problem, the impression of a conversation with a real person will be very important. Research shows that 56% of respondents indicate that they gave up chatting with a chatbot because their problem was too complicated. In the case of an AI based chatbot, the customer may not know that they are having a conversation with a bot. Even if the conversation goes beyond the level of the bot’s knowledge, a consultant may join in at any time and provide an appropriate answer, teaching the bot how to behave in a similar situation next time.
A process-based chatbot will also provide answers to questions, but only to those that the user chooses from several displayed variants. If these are simple, frequently repeated questions or requests, customers will be grateful that they don’t have to type them in and can simply choose one of the options offered.
Size of Customer Service department
If your Customer Service department is large enough to have its own customer communication platform, a process chatbot may not be sufficient.
Platforms for building simpler bots do not provide the features that help the consultant in continuous handling of multiple queries at the same time, particularly if the nature of the requests is complicated and the consultant will have to join the conversation often.
When implementing an AI based chatbot, the supplier also implements an appropriate platform for communication with customers. It is adapted to handle the bot and multiple queries from different channels simultaneously.
The customer persona
Who your chatbot will talk to is important for several reasons. Customer persona analysis will not only help you to build your bot (e.g. to select the tone of messages), but will also indicate which implementation strategy to choose.
If the chatbot user is, for example, a sales representative (or de facto your employee), then the nature of his or her submissions to your Customer Service department will be more complicated. In such a case, it is better to implement a AI based chatbot, which is more corporate in nature and is able to answer complex questions in context.
The age and level of technological advancement of your customers is also important. Almost 67% of respondents over 65 years of age indicated that a chatbot is another obstacle before talking to actual people, so in their case a rule based chatbot system will not contribute to increased satisfaction.
If you run a B2C business and your chatbot user is your customer or consumer, the nature of their queries will be less complicated. If you define a few frequently asked questions and clearly specify what your chatbot can do for a customer, a rule based bot will bring a lot of value to your customer service – especially if the average age of the customer is within the range of 18-34.
Specificity of the industry
Users have their own preferences when it comes to chatting with a bot, depending on the specificity of the industry. In case of requests concerning e.g. banking, finance or health care, users prefer to talk immediately with a human than a bot. This makes them feel more confident that their problems will be dealt with quickly and properly. In such a situation, quick involvement of a human in the conversation is crucial.
However, if your chatbot is supposed to automate simple tasks typical of your industry, such as checking the status of an order or answering simple questions, your customers will even expect a bot to answer them.
Many Customer Service departments have already gone through several attempts to automate communication. Employees are tired of implementing new systems which instead of making things easier, only complicate them.
The implementation of a bot that „learns from its mistakes” and is systematically developed by consultants is a unique investment because it becomes more and more effective with time. It’s like hiring a new employee who gains new knowledge every day but works 24 hours a day and never takes time off. Let’s add that the natural language processing technology is constantly evolving and there already are solutions available on the market that offer voice support for customer service. Building a knowledge base for an AI chatbot bot now will be crucial for the implementation of a voice version in the future.
However, there are many businesses, especially smaller ones, where such a large and complicated implementation is not necessary, and customers will appreciate the convenience of choosing the next available options in a rule based chatbot, even at the expense of its limited functionality.
When choosing the right strategy, you should always be guided by the preferences of your customers. Steve Cannon, CEO of Mercedes Benz USA said in an interview: “Customer Experience is the new marketing”. Implementation of a chatbot that actually helps your customers will bring you competitive advantage and ensure more dynamic development of your business.