Getting started with Salesforce’s very own AI Chatbot – Einstein Bot

Einstein bots are powered by Salesforce’s very own AI – Einstein and you can build these Bots without writing even a single line of code.

Chatbots are available to customers around-the-clock on their preferred communication channels, wherever they are:

• Messaging platforms like Facebook Messenger, WhatsApp, and SMS and Chat Channels like Websites and App.
• Smart speakers and smart home devices
• Mobile phones and tablets
• Desktops

Einstein Bots are built to support the human agent (and not to replace them) so that by the time agent takes over the case agent already has a good context of the situation and the necessary information like customer name, customer number, order number, etc. We don’t really need human agents for the simple and common use cases like tracking the status of orders, canceling the order, etc. Ideally, agents should focus on complex queries and this is where chatbots come into the picture. 

Chatbots can reduce the amount of work support agents do and free them from doing repetitive tasks. Requests like “What is the status of my support ticket” or “I am not able to log in” can easily be answered/processed by a chatbot without involving a human. 

Reach out to us for a quick chat: Salesforce Einstein Chat BOT

8 steps to get started with Einstein  Chat Bot

1. Prerequisites for Building a Bot

• You must have a Service Cloud license and a Live Agent license
• Omni Channel must be set up (Presence configuration, Routing configuration & Service Channel for Live Agent)
• A Live Agent Deployment
• Live Agent Chat button set to routing type: Omni-Channel
• A Queue for your Chatbot agents
• The agents must be Service Cloud users
• An Embedded Service Deployment (previously known as Snap-in Deployment) to host the Einstein Bot

2. Make a List of Jobs Your Bot Will Perform

This is where you pick the appropriate use cases you want to handle and automate using a Bot. 

Start by identifying the top intent your customers might have when they are on a website page/landing page/community page / App.  For example, to get help with login issues, to ask general questions, to know the status of order, etc.

3. Form a Team to Build Your Bot

It’s crucial to define who your stakeholders are and the role each one will play in developing and maintaining the bot.

1. Executive Stakeholders – who determine the bot use cases based on the business strategy and bot growth plan
2. Subject Matter Expert – Who delivers the insights and content that the bot uses to interact with customers?
3. Admins or developers, who build the bot in the Bot Builder. Contact us today if you need us to build the Bot for you

Your Bot team might include Product Managers, Salesforce Administrators, Developers, and others.

4. Decide the Type of Bot You’ll Need

Now it’s time to decide which type of bot is best suited for your business needs. 

Menu-based Bots: These are easy to set up and allow users to choose from several options, presented in the form of menus or buttons. Useful to increase conversions, engagement, and lead generation.

AI-Powered Bots: These types of Bots are trained to have human-like conversations using a process known as natural language processing (NLP). AI-Powered Bots learn from your customer’s input, understand context & respond accordingly. The chatbot continuously learns & improves to provide the most accurate answer every time.

Hybrid Bots:  You can create a blended experience with Hybrid Bots which use menu-based methods and NLP methods as and when required.

5. Form a Team to Build Your Bot

A chatbot flow is a structure that determines how a chatbot conversation will take place called “conversation design”, taking into account the questions your chatbot would ask and the various replies that a user could provide.

You can create a chatbot conversation flow diagram using tools like draw.io, whimsical, etc.

First, create a high-level chatbot flow diagram. For example,

Knowing the primary objectives of your Conversational Interface is an essential first step. Discuss with stakeholders and create a final conversation flow diagram which will have a final bot-message text copy (maintaining the brand tone) that will be shown to the user during chat conversation and actions that need to be performed in the background e.g. Create new lead or case, update a lead record, fetch case information by performing a search in Salesforce CRM. For example,

A chatbot’s content must be chosen before it can be built. It’s best to start with the audience.:

• Who is the chatbot for?
• What goal or purpose does the chatbot serve?
• Does the name and personality of the chatbot match the business’s brand?

The answers determine the content. Perhaps the people best suited to write content for chatbots are writers.

Topics Mapping:

• Will your robot have a niche or be more versatile? Will it cover a broad variety of issues or concentrate on a select few?
• Which inquiries it will be able to comprehend and which responses it will be able to offer

Then decide on Bot personality and tone of voice:

• Your bot should speak the same language your target users speak and its overall personality should reflect the brand you’re working for.
• Choose if your bot should be formal or informal, serious or fun, verbose or concise, etc.

The content is decided by the responses. Perhaps writers are the ideal candidates to create content for chatbots.

Learn as much as you can about the users your chatbot is targeting:

What is Einstein Bot made up of?

Message

Display a message to your customer. You can also use merge syntax to display collected information to your customer. For example, “Your order number is {!variableName} or use dot notation to access properties in variable of type Object “Your email is {!variableName.email}.

Question

Ask your customers a question and save the answer in the variable

Action
Perform action on your data such as update or get data and store it in your variable

• Apex – Execute flow and transaction control statements on Salesforce 
• External Service – <Write one liner description here>
• Flow – Collect data and performs actions in your Salesforce org or an external system
• Generate a completed goal log – <Write one liner description here>
• Generate Knowledge feedback log – <Write one liner description here>
• Object Search – <Write one liner description here>
• Send email – <Write one liner description here>

Rules

Perform the following specific actions:

• Call a dialog from within the current dialog
• Redirect to a different dialog
• Clear a variable value
• Transfer to an agent
• End Chat
• Send a message
• Set conversation language
• Set routing type
• Set variable

Entities
Entities store the type of data that you want to store

Extend Bot capabilities with – 

• Apex
  Among all of the action types, Apex has the broadest use cases.
  <Mention some use cases here>

• APIs
  Add an action from a registered external service in Bot Builder to allow your bot to send and receive data between Salesforce and external API providers.
  <Mention some use cases here>
  
• Einstein Intent
  You can make the bot smarter, and more conversational using Natural Language Processing (NLP). You can start building your intent model. Intent models are made up of multiple intents, which have many utterances assigned to each intent.
  <Mention some use cases here>

Read our blog – Best practices for building Einstein Bot for the best practices

6. Form a Team to Build Your Bot

Examining the historical chat transcripts will tell you if your chatbot needs natural language processing (NLP).  If visitors are opting to free text their concerns instead of selecting pre-planned menu options, then it’s important to add NLP in your chatbot.

You have to provide example sentences, known as “utterances”, to the bot and categorize these sentences with a label, known as “intent”. 

• Download historical chat transcript data to extract real utterances used by customers in the chat conversation and upload these scrubbed utterances to the bot configuration through the data loader. 
• You can also use an utterance generator. It generates multiple variations to be used as sample utterances for your intent.

7. How do you measure your Bot’s success

We recommend keeping the list to three KPIs to start

• Number of cases deflected
• Customer feedback ratings
• Leads generated from Bot
• New enrollments within the bot
• Product purchases within the bot

8. Go live, analyze and iterate!

You are ready to launch Einstein Bot into production once your conversational flows have been developed and extensively tested.

However, the project is not yet complete. Instead, this marks the beginning of a brand-new phase: dialogue analysis! Knowing how your chatbot is operating will help you determine whether your design decisions were sound or whether there is room for improvement.

Conclusion

Now that you have learned about the power of the Einstein Bot, it’s time to get started! 

Start your Salesforce Einstein Bot journey and enhance your customer’s experience, increase the efficiency of your customer service with the 24/7 availability, scalability (processes hundreds of concurrent requests at once) and affordability of chatbots which can save your time and money.

Thinqloud is a trusted Salesforce partner. Learn more about us here to see how we can help you.

Reach out to us for a quick chat: Salesforce Einstein Chat BOT