Have you ever wondered if your Bot is “intelligent” enough to handle all possible conversations with the end user? What if the Bot needs to transition the conversation to a live human agent to ensure maximum end user satisfaction? The Bot should be smart enough to recognize when it needs to hand-off the conversation to a human agent and ensure that the end user experiences a smooth transition from the Bot to the human agent.
The following are some of the possible scenarios that would need Human intervention:
The Service desk Bot may be trained to help the end user with troubleshooting most of their mundane issues. But what if the end-user’s issue is complex and the Bot is not able to provide anymore troubleshooting? Then the Bot will need to involve a human agent to help the end user with the complex issue. The Bot must be smart enough to suggest if the end user would like to “Talk to a Human Agent” to resolve the issue at hand.
Related: How Can AI Bots Increase IT Helpdesk Support Efficiency?
Sometimes, the end user may be in a rush and may not want to wait for the Bot to help with his issue. The user should be given a choice to directly talk to a human agent instead to conversing with the Bot first. This can be provisioned with the help of menu options.
One of the menu options of the Service Desk Bot should be “Talk to a human agent”. This would allow the user to directly connect with a live agent in case he prefers the human agent over the Bot for assistance.
There can be scenarios where the user may be frustrated/unhappy with the assistance provided by the chatbot. Thanks to technologies like natural language understanding and sentiment analysis, the Bot can be trained to infer the “mood” of the user by analyzing the content of the conversation.
If the Bot “senses” frustration, the Bot can suggest “Chat with human agent” as an option to the end user to let the user decide if he would like to continue the conversation with the Bot or rather speak with a live agent
Based on the business jargon, the Bot can be trained to identify “critical” situations. For example, in case of an insurance company, if the end user is looking for some critical information regarding the policy claims for a deceased family member, the Bot can be trained to directly involve a human agent to handle such conversations. This ensures maximum end user satisfaction.
In some scenarios, the live agent may want to “observe” the performance of the chatbot. They keep a close watch on the conversation of the Bot with the end user and in case, the Bot needs to suggest a critical operation to the end user, the bot can confirm the action with the agent before suggesting it to the end user.
Now that we have looked at some of the possible scenarios for a human hand-off, how exactly should the Bot handle the transition?
Do you remember the frustration when you are on a customer service call and you have already shared all the information with an automated virtual voice agent and then you are asked to repeat the entire information when the live agent connects? We need to ensure that we provision a seamless transition to ensure the best user experience for our end users.
Once the Bot needs to transition the conversation to a live agent, it needs to ensure the following:
Smooth transition to live agent
Conversation history to be shared with the agent
Smooth disconnect from the agent
First and foremost, the Bot will need to “acknowledge” to the user that the conversation will now be transitioned to a live agent. Then the Bot moves the end user’s conversation to a “wait” state and check for an available human agent. The Bot can also provide the end user an option to confirm if they would like to “wait” for an available agent or try again later. This would help the end user disconnect, in case the “waiting” time seems long.
In case the end user decides to “wait” for an agent and then types in a query while “waiting” for the agent, the Bot will respond “Please wait while we connect you with the next available agent.” The Bot will not handle any incoming queries from the end user while the conversation is in “wait” state.
The Bot design will define how the waiting queue will be handled. Typically, Bots are programmed to handle the “waiting” requests in “first in first out” manner. Once a live agent is available, the Bot will intimate the agent that an end user is waiting for assistance. In case the agent accepts, the end user will be notified that he is now connected to the agent.
Once the agent and end-user are “connected”, the Bot will share the entire conversation history of the current chat with the end user either in the form of text messages in the agent’s chat window or the chat script can be send as an attachment via email to the agent. This will avoid the agent from asking any repetitive information that the Bot has already collected from the end user.
The Bot will continue to act as a “behind the scenes facilitator” and route the messages between the agent and the end user, thus creating the illusion that the end user is communicating directly with the agent and vice versa. However, all the queries from the end user are routed by the bot to the agent and all the responses from the agent are routed to the end user via the Bot.
On completing the assistance to the end-user, the live agent will have the provision to “disconnect” from the end user. The end user will be notified that he is no longer connected to the live agent.
Any query from the end user will now be handled by the chatbot and the agent will no longer be able to “view” any conversation with the end user.
Since the Bot “facilitates” the conversation between the agent and end user, the Bot can provision a richer user experience to the end user by providing multilingual support.
Imagine the scenario, wherein your service desk agents can communicate only in “English”, but your end users are spread across geographies and would prefer to communicate in their native language.
By leveraging the language translation services, the Bot can “translate” the message while routing it between the end user and the agent.
For example, if the end user asked a query in “Dutch” language, the Bot will translate the message to “English” before it is routed to the live agent. The agent would then respond to the query in “English”, which would be translated to “Dutch” by the Bot and then routed to the end user.
Thus the “human handoff” functionality with multilingual support will help you assist your end users across geographies and enhance their satisfaction levels with human intervention when needed.
If you’d like to learn more about this topic, please feel free to get in touch with one of our experts for a personalized consultation.