Topic: CRM

Leveraging AI with Salesforce Einstein Chatbots

Author: Scott Grant

Are you ready for an Artificial Intelligence Chatbot?

In 2018 Gartner predicted that Chatbots will power 85% of all customer service interactions by 2020. It’s 2020 now and they were probably overstating the rapidity of AI Chatbot adoption.  But they weren’t that far off the mark. Chatbot technology has made rapid advances and our understanding of what makes a good Chatbot has also grown. Now that our know-how and the technology are ripe – should your company start implementing AI Chatbots? We’ll look at Salesforce’s Einstein Chatbots to see if we can answer that question.

What is a Chatbot?

A Chatbot is an application that simulates human conversation via text message. It allows your customer to talk to a computer that then directs the customer to the resources that they request.

Chatbots allow you to add an additional channel of communication to your sales and support strategies.  You may have encountered Chatbots that did not respond well to your questions, and you may have encountered Chatbots that you practically didn’t know were Chatbots. The difference between these two experiences is Artificial Intelligence and Natural Language Understanding (NLU). Natural Languages are those languages that derived from humans – like Spanish, English & Japanese. Constructed Languages are artificially made like Esperanto or Java. Chatbots that support natural language processing seem more responsive and human even when we know they are a computer. Chatbots with Natural Language Understanding are “smart” and provide a richer customer experience than those that do not.

Salesforce’s Artificial Intelligence Platform, Einstein, has Bots that use Natural Language Understanding and that plug into the Service Cloud and Live Chat features.

Why would you want one?

Customers like answers to their questions quickly. Maybe your customers are still filling out a web form or emailing support@mycompany.com and you’re looking for a way to be more responsive. Maybe you’ve recently implemented Digital Engagement and your customers are getting immediate responses from your agents from Facebook, SMS texting and Web Chat; great for your customers but your agents are busier than ever.

“By 2021, 15% of all customer service interactions will be completely handled by AI, an increase of 400% from 2017,” says Olive Huang, research vice president at Gartner.

Enter Einstein Chatbots to help relieve your agent load. And if your customer needs to speak with an Agent, the Chatbot acts like a concierge and makes the connect between them.

How would you approach an implementation?

Even though Einstein Chatbots are intelligent, that doesn’t mean you turn them on and let them rip. Planning is essential for having a good Chatbot Customer Experience (CX).

If you are planning a support Chatbot it’s important to ask the Service Desk about common questions that come up that the Chatbot can handle.

          Questions like:

          “How do I change my password?”  

          “What is the status of my order?”

          “How do I print?”

          “Are there stores nearby me?”

Just as you’ve asked your support folks for help in identifying the types of questions that the Bot can answer, it’s important to have people who work with language (otherwise known as Writers!) craft the dialog that the Bot will use to talk to your customers.

Now you know what you want your Bot to do and what it will say.

What would your Chatbot look like?

Does your Chatbot have a personality? Does it use Emojis? What font does it use? Does it have a persistent menu that a person can choose?

Let’s say that Order Status is 14% of all your customer inquiries. We want to save agents from getting a name, an order number and searching for the order. Instead we want our Chatbot to ask the customer for their order number, look up the order and return the status field to the Customer and then give them the option to talk to an agent.

First, we design a chat menu – we’re going to have an option for Order Status, Appointments, Search and Transfer to an Agent. When the customer selects that Order Status, our Bot will ask for the Order Number, look up the Order and say in human language what the order status is. Then we’ll ask if they want to go to the Main Menu or be Transferred to an Agent.

Here’s our Bot, Jana, answering the customer’s request for a live chat.

And here it is responding to the Order Status request.

Finally, after looking up the order number it returns the information our customer needs.

Here are some more things to think about as you consider how to put Chatbots and AI to work for you.

When setting strategy for technology adoption, service leaders must remember to:

  • Manage expectations by setting modest goals for initial projects and by describing projects with accurate terms.
  • Expect, allow and even encourage, experimentation, but never deploy poor-quality solutions to users.  
  • Ensure that Chatbots have clear business rationales and objectives and are not just based on the “coolness” factor.
  • Factor in that while some technologies will work on their own, others will work best in combinations, such as a Chatbot in a mobile application, driven by a powerful natural language processing engine.
  • Genuine human relationships are an organization’s most powerful resource, Chatbots are an assist that will allow you to spend more quality time with your customers.

How can we help? 

Whether you already utilize Salesforce and are now considering an Einstein Bot or you are interested in implementing Salesforce for your organization, let us know! We’ve got a team of experts who would love to speak with you.