How to develop a chatbot in 5 steps

According to Gartner, by 2020 over 50% of all mid-sized and large companies will be using chatbots. Consumer acceptance for interacting with chatbots is also increasing. A 2017 survey by Statistica shows that by this time 34% of consumers surveyed were ready to communicate with a chatbot instead of a customer service employee. But chatbot projects in companies do not always lead to success. A thorough analysis and planning is indispensable for chatbot development. In this article I will guide you through the 5 steps that are necessary for the development of a chatbot and the deployment in your company.

Chatbots with human qualities are going to handle the contact between consumers and companies more and more often in the coming years. There are many different variants and applications of bots. Chatbots can be used, for example, as a work planner to ask for missing information such as the customer’s order number or e-mail address. But they can also act more in the background and provide information to the customer service staff. Due to the wide range of applications for the virtual assistants, it is essential to determine right from the start, which role chatbots and artificial intelligence should play for all business areas.

The development of chatbots and their influence on services, marketing and sales will become increasingly important in the near future. In this article we explain the most important steps in developing chatbots for your business.

One of the most important steps in the development of chatbots is the definition of their tasks. What would you like to use the chatbot for? Which problems and challenges can a chatbot solve for your company? Where can a chatbot help your customers better? Think of opening hours, passing on meter readings or passing on product information. Chatbots in the customer service area are mainly used for three purposes.

1. As an assistant to the customer service employees, where the chatbot provides suggestions for answers. The service staff then only has to confirm or decline the suggested response.
2. As a work planner, where the chatbot requests necessary information from the customer in advance, such as e-mail address and/or customer number, and then provides this information to the customer service employee.
3. A chatbot that independently answers service questions.

To get started with a chatbot, we recommend that you first select only one of these tasks. Starting with just one task, ensures that you can measure the success of the chatbot, such as its efficiency and the increase in customer satisfaction.

Before a chatbot can be presented to customers, it must learn to generate appropriate responses that meet the right tone of voice when communicating with customers. This is the basis for a uniform and authentic communication of your company to the outside world. Answers can be generated based on examples from the past. In this way, the stored conversations between customer service employees and customers serve as learning material for the chatbot. However, the chatbot should not only learn on the basis of historical data, but also by giving input yourself.

As a company, you know best which image you want to convey to your customers. Enter your own predefined answers and the chatbot will do exactly what you want!

Another important point is the definition of business rules. Teach your chatbot where he can get information from, when he needs to access external systems and when he needs to request additional information from the customer. In other words: what is the bot allowed to do and what is it not allowed to do? Chatbots are currently becoming more intelligent and able to learn autonomously. But it’s important to set business rules to keep control.

Examples of business rules include opening hours. When a customer asks how long your company is open today, the chatbot first needs to know what day it is today. He must also have the daily opening hours and any deviations from them “in mind”. If a customer has a question about an order, his customer number is needed. As soon as the customer number is not available, it is the bots job to ask for it. Only then, he can proceed to the next step. The business rules are the navigation system for your chatbot. These rules control the chatbot so that he provides the right service.

A chatbot must learn to understand what you mean. While in the past chatbots were mainly controlled by a script, nowadays bots can be made smarter using artificial intelligence and technology. This allows them to recognize the intention of the post (intention), but also to filter data from the message (entities), such as an email address or a customer number. In this way a bot can be trained to understand messages.

To train your chatbot, you can use your existing customer service system. To do this, categorize the existing customer inquiries and answers. In this way, you ‘feed’ the chatbot with various posts and various possible answers to which the customer could ask a question.

A connection to an external system provides a 360-degree image of the customer and can help the chatbot perform certain actions better. Such a link to external systems is not absolutely necessary, but it will considerably expand the possibilities of your chatbot. A connection to your CRM tool, with more information about your customers, can be useful. You may even need to address a specific target group differently. This differentiation can then be applied.

It can also benefit the usability of the chatbot, if it receives information from external systems to answer the customer’s questions. For example, a chatbot can check if there is a malfunction, so it notifies the relevant employees via SMS. Or it can send an instruction manual to a customer by e-mail when the customer asks for instructions. Finally, the chatbot notes all customers with whom he has had contact and has their data at his disposal.

Test your chatbot continuously. Does it always work equally well? And if not, why is that? Can you feed him with more examples to improve his service? Regular tests will give you a better understanding how the chatbot performs. Take the time to take this step, because the success of your chatbot has a direct impact on customer satisfaction.

Ready for take off…
The last step you need to take is to get your chatbot up and running. It doesn’t matter which channel you choose, because the chatbot can be used everywhere: from Live Chat, WhatsApp to Facebook Messenger to Twitter DM. Ideally, you should use your chatbot in the same environment in which your customer service employees work.

Within our company, the chatbot takes care of first line customer requests. This allows the service team to focus on the remaining, more complex customer issues. A big plus point: employee satisfaction increases!

Think big, act small
Would you like implement a chatbot? Start with a smaller application so that you can easily measure and evaluate its usage. The steps discussed in this article describe a repetitive process in which you are constantly aiming to improve the chatbot service. Don’t forget that you have a lot of knowledge of your own. A chatbot is therefore an extension of your service and your company.