Chatbot — Knowledge base

Rohan
Rohan
Nov 4 · 4 min read

Problem Statement

Imagine you’re into Sales and on your company’s sales portal, where you were being Greeted by a friendly chatbot at the bottom of the screen. But when you ask it a question, you’re disappointed by the following response — “Sorry! I don’t Understand.” Isn’t that Frustrating!

Unfortunately, its a common experience as companies eagerly adopt chatbots to improve the users experience on respective application. In sales it gets more critical because only quality which should be en-cashed from sales is to make them more & more “Productive”.

Solution

The best sales chatbots are able to seamlessly handle various types of interactions, like — finding qualified leads, prospecting, know about sales flows to better prioritize actions and accordingly setting up tasks and events and achieving them to Close eligible leads. Along with that answering Frequently asked questions (FAQ’s). They also do it in a natural way as if they are interacting with a Personal Assistant.

Challenge — How does a chatbot know what information to provide when a user asks a question? And what about answering common questions about the a Lead or Company?

That’s where Knowledge base comes in. A knowledge base provides your chatbot with the information it needs to accurately respond to user’s questions.

Think of a knowledge base as a collection or repository of information about your domain or organization that is maintained by you and deployed by your chatbot at just the right time: when it’s requested.

Let’s take a look at the key knowledge base elements you should be aware of when planning how to deploy and maintain your chatbot.

FAQ Topic: What questions my chatbot should answer ?

You need to add those questions which you anticipate them asking. There is no single list of topics that can account for everything users (sales) wants to know about. You’ll need to identify the most common questions asked by sales teams.

FAQ response: How do I want to answer this question ?

Each FAQ you add will require a response, which the chatbot delivers to the users when they ask a related question. Responses could be a pure text reply or a link to additional information that’s too lengthy to include within the chatbot window. You could also include a video, which would be great for questions around company culture or showcasing employee testimonials

FAQ Related Phrases: How do I account for users asking the same question in multiple ways?

Since there are multiple ways to ask the same question, related phrases will serve as the training data for your FAQ topics. These don’t have to be questions necessarily. They can be short phrases or even individual keywords

Unanswered Questions: What are missing ?

This is a very important piece of the knowledge base. One major benefit of having the chatbot connected to a knowledge base is its ability to store and surface topics that you missed at deployment. When a user asks a question that the chatbot is unable to answer, that question is flagged as unanswered. As you review your knowledge base, you’ll be able to add them as new questions your chatbot can answer.

Some of the FAQ’s like -

Question 1 — What is the salesforce version I am using currently in ? Answer “Its Enterprise Edition” OR “You could find the version details here

Question 2 — What are various status an Approval Flow goes in its life cycle ? Answer“An approval request could have various status in its life cycle as Not Submitted, On-Hold, Approved, Rejected, Notified, Submitted, Assigned, Re-assigned, Approval Required, Pending Approval, Approved, Rejected, Cancelled.”

Question 3 — How can I quickly create a quote from MAX ? Answer “Sure, I can create one for you. Shall we do it now ?” {And direct to relevant conversations}

Question 4 — Hey, I am new here. Do you have any reference to understand Various Company Policies. Answer — We could collaborate with companies and get some reference like videos or links which could be used to answer such question.

Pros and Cons

Pros

  • Makes user more comfortable with your chatbot.
  • Scalable by including Unanswered questions or intents in your chatbot.
  • Gels well with existing chatbot framework
  • Categorization of FAQ’s like OnCompany, OnApprovals, OnQuote, OnSaleforcePortal, to ensure its maintainable
  • FAQ’s approach could help to collaborate more with companies and such features could help org. to spread voice with the help of chatbots and gives a sense of personnel to resp. org.

Cons

  • Anticipating and building repositories could be tasking. Till the time you start building one.
  • Links being upgraded or closed — The links shared in the response are updated. Those needs to reflect in Chatbots response as well. Mitigation could be — Automating the test case and get such pieces resolved before deployments.
  • Data Privacy — Information i.e. the collections of information we are sharing to user are private in nature. So need to be cautious while deciding the answers. Mitigation could be — Collaborate and know the policies beforehand.
  • Role based access control — If the answer include a link which user dosen;t have access to it again brings frustrations to user. Even its not in chatbots control. Mitigation could be — Collaborate and know the access control policies and include similar nature in respective chatbot.

Rohan

Written by

Rohan

Handeling situation with zeal and passion

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade