Hybrid chatbot: how to make humans and robots work together

Adrien Lemaire
Wed, 04/04/2018 - 18:27

Customers' expectations towards businesses have been impacted by a change of how customers consume. 64% of consumers now expect a brand to interact with them in real time, and 80% say receiving immediate responses positively influences brand loyalty. Channels such as messaging, used by 2 billion people around the world, are a way to adapt to these new expectations.

However, meeting these new expectations while maintaining quality customer care is a major issue for companies. To face the growing volume of messages on multiple digital channels, centralizing customer care management becomes essential. By answering through a single platform, activity across different touch points can be optimized, agents' training time is reduced, and piloting is done more easily.

Recent advances in natural language processing have contributed to the evolution of chatbots that automate answers to customer requests. Yet the answers given by these classic chatbots is partial and disappointing. In fact, on average, more than 50% of client requests are not recognized by chatbots. This forces redirecting to an agent in more than one case out of two. A majority of users are misunderstood due to how they formulate their request, resulting in bad customer experience and lower satisfaction.

The limitations of this technology show that it is not possible to replace agents with chatbots. It is better to use each resource for what it does best and set up hybrid chatbots.

1. The solution: A hybrid chatbot

A hybrid chatbot can be set up on different types of channels, in collaboration with an agent:

  • Live Chat: Live Chat (or Web Chat) is a tool allowing visitors of a website to talk in real time with an agent. The customer can access it from a window appearing on the company's website. The advantage of chat is that it assists customers in real time.
  • Messaging: Messaging includes applications such as Messenger, WhatsApp, WeChat and Viber, which allow you to exchange at a "fast asynchronous" pace.
  • In-app messaging: this system allows communication between a company and its customers within its mobile application. The in-app messaging interface looks like an application such as Messenger or WhatsApp, but is integrated into the company's mobile app.

A hybrid chatbot is a program that dialogues with humans to provide an automated and personalized reply. It intervenes only through an instant messaging channel in addition to the agents (live chat, messaging or social networks).

A company that already uses a digital customer interaction platform simply adds the hybrid chatbot as a plugin: a "chatbot" agent is added. It will be able to instantly read all incoming inquiries and answer when he is able to.

There is no change in the agents' job or customer replies process, the hybrid chatbot acts as a new virtual agent.

A hybrid chatbot can take advantage of the existence of agents on the same communication channel to answer only the requests that it is certain to have understood. And this completely changes the perception of the customer who dialogs with bots only when they understand their request.

2. The results obtained by combining Dimelo and Botmind

Setting up a hybrid chatbot with Dimelo and Botmind is very easy. The hybrid chatbot is added as a new agent in the system. This agent receives all incoming requests and depending on whether the client request is understood or not, it initiates a dialogue or recategorizes the conversation.

There is no change in the agents' processes. The only notable change will be the reduction in the number of recurring requests to process because the virtual agent will process them. During a possible bot redirection, the agent accesses the history of the conversation between the user and the hybrid bot, in order to have all the information he needs.

By performing this type of integration, Dimelo and Botmind were able to observe the beneficial results on average response time and customer satisfaction:

53%
On average, the Botmind hybrid chatbot is capable of responding to 53% of customer requests for frequent intent. Agents see their workload reduced by more than half in terms of answering frequently asked questions.


28 seconds
On average, the response time to customer requests drops by 28 seconds thanks to the hybrid chatbot.


26%
Customer satisfaction increases by an average of 26%.

In order for the real revolution of Artificial Intelligence to occur, it is essential to combine it with human intelligence.

For Customer Care, the hybrid chatbot is an excellent example of a relevant use of AI. A program is used for what it does better than a human: responding to many simple requests instantly. A human intervenes for what he knows better than a program: to understand the complex problems of another human being.

About the author

Profile picture for user adrien
Adrien Lemaire Content Manager

Subscribe to Dimelo Newsletter!

Hubspot tracking code

Dimelo Newsletter

Regularly receive news and trends about digital Customer Care

Any question?

Our team replies on Facebook Messenger!