AI for Business & Productivity · Intermediate level

Intermediate level · Module 3: AI for Customer Engagement · Lesson 1 of 3

Scoping a Customer Chatbot

Decide what a chatbot should and should not do, and build its knowledge base.

In this lesson you will

  • Define a chatbot's scope and tone
  • Create a knowledge base of verified answers

Key concepts

use casesout-of-scope topicsknowledge basedisclosure as automated

The lesson

Chatbots work best for frequent, simple questions: opening hours, prices, delivery areas, how to order. They are poor at complaints, refunds and anything needing judgement.

Build a knowledge base of verified answers from the business's real information and update it whenever prices or policies change.

The chatbot must say it is automated, keep a friendly tone and never invent answers outside its knowledge base.

Step by step

  1. List the 20 most common customer questions.
  2. Mark which the chatbot should answer.
  3. Write verified answers for the knowledge base.
  4. Write the chatbot's introduction message.
Case study

A Juba phone shop's first chatbot

A composite, illustrative case: a chatbot that helped once its scope was narrowed and escalation was clear.

Context

A phone and accessories shop in Juba (a composite example) received dozens of WhatsApp messages a day asking about prices, stock, opening hours and repairs. Two staff members answered between serving customers.

Challenge

A first chatbot tried to answer everything. It quoted outdated prices, promised repair times the technician could not meet and frustrated customers with complaints.

What they did

  1. Narrowed the scope to opening hours, location, payment options, stock categories and how to book a repair.
  2. Built a knowledge base updated every Monday with current prices, which the bot quoted only with the date of the price list.
  3. Added escalation: complaints, refunds and repair timing questions went straight to staff, and the bot said when they would reply.
  4. Reviewed ten conversations every Friday and fixed wrong or missing answers.

Results

  • Staff spent less time on repetitive questions and more on sales and repairs.
  • Complaints about wrong information stopped once prices came only from the weekly list.

Lessons for your own work

  • A narrow chatbot that answers simple questions correctly is better than a broad one that guesses.
  • Clear escalation and weekly review are what keep customer trust.

Practice activity

Write the chatbot specification and knowledge base.

Expected result: A clear scope and a verified knowledge base.

Check your understanding

  1. Which questions suit chatbots best?

    Show answer

    Frequent, simple questions such as hours, prices and how to order.

  2. What must a chatbot tell customers?

    Show answer

    That it is automated.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.