Improve Your User’s Service Experience with Relevant Routing

Have you ever run into this scenario: You’re talking in a customer service chat inside a website. You ask to speak to a human and are routed to an agent who has zero understanding of what you need. And by the time the agent directs you to someone who actually knows what you want, you’re already frustrated with the entire experience.

The thing is, you’ve asked to speak to a human because you need to talk to someone who knows how to solve your problems. So it really does you no good if you get directed to someone who doesn’t have a clue of what you’re talking about. Can we just skip the part where the first “expert” directs you to a second, relevant expert already? What you need is to speak with an agent in the right department, immediately! Someone whose expertise is exactly what you’re looking for.

With Watson Assistant you can provide your users with exactly that sort of service experience! Watson Assistant’s Zendesk and Salesforce integrations automatically detect when and why your users need a live agent, and seamlessly transfers them to agents best suited to meet their needs.

So let’s go over an example to see how you can build such an experience for your users.

Take a look at the assistant below. This assistant has been designed to transfer users to human agents based on the specific banking issues they’re trying to resolve. Here is a user who asks to speak with a human after retrieving her account summary:

The user needs to speak with a human after asking about her bank account.

And here is a user who needs help withdrawing money from her account:

The user needs to talk to an agent after trying to withdraw some money.

Notice how in these examples:

• The assistant begins transferring the user the moment they need a live agent. Even though the assistant knows when to request for an agent, it’s still up to the user to confirm that she wants to escalate the conversation and talk to a human.
• The user did not have to choose which department she would like to speak with. Not only does the conversation identify which department of agents to direct your conversation to, but your users will also be notified if no agents are online to speak.
• The agent picks up right where the assistant left off and is able to help the user immediately. This is not only because the service agent knows which department the user is directed to, but the service agent also receives a quick summary of what the assistant conversation was about.

So how does all of this work? To create your own seamless transfer experiences, you only need to go into your Watson Assistant’s dialog builder. Specifically, inside your dialog nodes that use the Connect to Human Agent response type:

A Connect to Human Agent response type node that handles user’s requests about their account information.

A second Connect to Human Agent response type node that handles user’s requests to withdraw money.

Notice that the Connect to Human Agent response type has four fields that allow you to customize your transfer experience:

• The first two fields, “Response when agents are online” and “Response when no agents are online”, specify how your assistant responds to the user. The first message is what your assistant will respond with when agents are online and when the user is ready to request an agent. The second is the message when there are no agents online to service the user.
• The “Message to human agent” field is the message your assistant sends to the human service agent. This message helps the agent know what the user needs help with.
• Finally, the “Service desk routing” field is the department of agents that the user should be transferred to. When your conversation reaches this Connect to Human Agent dialog node, it will be transferred to whatever department is specified. If no agents in the department are online, or if the department does not exist, your user will fall back to speaking with any online agent.

As you can see from the Simple Banking Assistant’s dialog builder, you should design the conversation tree to distinguish every use case from each other. Each use case can get its own Connect to Human Agent node, and that will allow you to customize how to respond and cater to that scenario.

So as you build your own conversation trees, remember this: Keep your dialog tree clean by dividing each path into conversational use cases. And don’t forget to fill in all your Connect to Human Agent nodes. Make each message and each department specific to what your users need!

Setting up integrations with Salesforce and Zendesk only takes minutes, and now with little extra effort, you can perfect your users’ transfer experience.

To learn more about setting up these integrations, check out our how-to videos for Zendesk and Salesforce. If you are not yet a Watson Assistant customer but you’re ready to start building, sign up for Watson Assistant here.