Fishing for Einstein Chatbots
Chatbots are a great way to engage students and other constituents, but a lot of schools have trouble figuring out how to get started with them. Some of the initial technical details like how many bots you need, where they should be located and what type of content they should provide can be a bit overwhelming. It may feel a bit like going fishing for the first time: Even though you love to eat fish, a lot goes into catching and preparing them that you may not know how to do. In much the same way, a lot goes into setting up a bot compared to using one. But with the right approach, configuring a bot can become just as straightforward as casting a line out to sea is for a seasoned fisherman.
A colleague and I were recently working with a customer on a Salesforce Advisor Link project that also involved standing up chatbots to help with student communications. I asked our customer some questions that I thought would help them start thinking about what they wanted their bot to do. I told them to think about things like menu items, topics, and the types of responses they wanted to be able to send to their students, assuming that this would set them up nicely for follow-up discussion. Sadly, that discussion ended with us wrapping up early without any answers.
The customer wasn’t happy, and my team and I weren’t very pleased either. That’s when I realized that I was the one who had made the mistake. I had been asking our customer to go fishing without showing them how to fish. And the questions that I was asking were the equivalent of asking the person who’s never fished what size hook they wanted to use and whether they preferred a float or a casting spoon. Everything made perfect sense to us but none of it make any sense at all to our customer.
So how did we finally get everything figured out? I started by writing out a sample user story that the customer could easily follow along with, similar to the one below:
As a student, I want to interact with a chatbot to get answers at any time of the day.
When I open the chatbot, on [this webpage], the bot displays this text “Hello, I’m the Cody-Bot here to help you. Please ask your question or choose from one of the following items:
- Financial Services
- Campus Life
- Advising
When I select the Financial Services item the bot displays a list of items to choose from
- Inquiry about a Bill
- Inquiry about payment plans
- Inquiry about a refund
When I select the Bill item the bot displays a list of my current bills
- Bill 1 details
- Bill 2 Details
- Show more…(if more are listed)
When I select the Payment Plans the bot displays
- Please click here to see about how to establish a Payment Plan (URL to XYZ portal site about setting up payment plans)
When I select the Refund the bot displays
- To see about your refund please call Financial Services at 555.555.1212
When I select the Campus Life item the bot displays a list of items to choose from
- Campus Housing
- Off Campus Housing
- Greek Life
When I type in the question: ‘What bills do I have?’ The system will display a list of my bills. Then when I click on one of them it will display Balance Due, Due Date, and Name of the bill.
When I type in the question: ‘How do I get on a payment plan?’ The system will display the text: Please click here to see about how to establish a Payment Plan (URL to external site about setting up payment plans)
When I type in the question: ….? The system will display the text: ….
After seeing this, our customer gave us the information we needed within a day and we were able to build a POC in less than a week. The user story had the same questions we had asked in our first meeting but the key difference is that instead of simply asking open ended questions, we showed them how to give us the answers we needed.
For example: “As a student, I want to interact with a chatbot to get answers at any time of the day about my financial questions.” gives us the answer to the question “Why do you want to have chatbots enabled at your institution?”
- “As a ….” tells us who is going to be using the bot.
- “I want to….” tells us why the student is using the bot.
Using this method helped us teach our customer how to fish.
If you’re thinking about going fishing for a Chatbot, use the outline above as a starting point and remember to start small. Just like in all fishing stories, that 30 pounder will come along but you’ll have to catch a few smaller ones in the beginning to familiarize yourself with the process. If you keep your scope down a single bot and 2 to 3 categories at first, you’ll see some early successes pretty quickly while also giving yourself room to grow.
Additional Resources
- Einstein Chatbots Best Practices
- Developer Cookbook
- Getting Started: Service Cloud: Einstein Bots (Accelerator Webinar)
About the Author
Peter Fife is fortunate to do what he loves while making the world a better place. He does this by helping Non-Profits and Education amplify their mission utilizing the Salesforce platform. He’s a Solution Architect who’s been with Salesforce.com’s Social Enterprise .Org for the last two years, but has been working on the platform since 2003. One of the proudest moments of his career was speaking on day 3 of Dreamforce in 2004, yeah that’s when there were only about 75 people left. This was about transitioning the grassroots user community he started on yahoo boards over to the supported Salesforce user community. He’s equally proud of being part of .Org and writing this post as part of Salesforce.
