Airline Customer Support via Chatbot

Airline customer support departments are often overloaded with numerous queries from their customers. Some of the most common queries include information about: refund status, critical information, flight schedule, boarding passes, lost baggage, or similar.

To reduce costs, provide instant support, and maintain a high level of quality of customer service, airlines are implementing chatbots as a solution that automates many customers' queries. Besides cutting costs, chatbots help airlines to offload customer support agents by giving them enough time to focus on complex, high-value queries.

On the other hand, chatbots can offer added-value services to customers, such as seat upgrades, extra baggage upgrades, and priority boarding that help increase revenue.

Nevertheless, customers sometimes need help and assistance from the agent. To keep customers satisfied, route them to agents and keep the full context of the conversation.

 

This use case will demonstrate how to define, design, build, and deploy a chatbot which can help your customers find and receive information related to their flights faster, or transfer them to an agent if the customer desires it.

 

PROCESS WORKFLOW

IMPLEMENTATION STEPS

Pre-requisites:

1. The first step before you begin designing your chatbot is to talk with your customer service/support team regarding the most common customers’ inquiries, in order to identify highest-volume customer intents regarding the topics you want to cover with a chatbot.

2. Based on the input from the customer support team, group the most common inquiries into intents and design the dialogs for the chatbot.

3. Create intent training files for each intent-based dialog using the most common phrases you identified.

Steps over Web Interface

1. Login to the web interface and select Answers module.

2. Click the Create chatbot button and create a new chatbot. Enter Bot name and Channel – select the one your customers use the most (WhatsApp for example). Before you activate the bot, make sure you also configure the rest of settings, such as the Sender number, Language, Escape phrases, and Session Timeout.

3. If you have the intent training files prepared, import them. Otherwise, go to Intent tab and manually create all intents required for the dialogs. Each intent needs to be trained with a minimum of 50 phrases but in the specific example of refunds, you can add all the possible variations of phrases your customers use when asking for a refund. This will enable the NLP engine (Natural Language Processing) to work better when recognizing intents.

4. Based on the inputs from your customer support team, create all required attributes. Attributes will help you with saving relevant information, e.g. user information, flight number, check-in, whether eligible for refund, etc. You can also define the attributes that will be visible to the agent if the conversation switches to that at some point.

5. Once most of the intents and attributes exist in the Answers platform, start with dialog design. Based on the conversation schema, use the required elements to build the dialogs starting with the selection of the intent for refunds for example.

6. Use elements like Send text, Attribute, Call API, Conditions to create dialogs according to your design.

7. For the Refund intent, start with Attribute element to notify the customer you need their flight number. After receiving the flight number and saving it, use the Call API element to find out the status of the flight. Through Conditions element branch the possible outcomes. Each outcome will have a different next step. For the cancelled flight, forward the customer to the dialog dealing with Refund policy using the Go To Dialog element.

8. As the customer is eligible for a refund, offer them options to either fill a web form with the necessary information, or send them to talk directly with an agent. If they select to talk with an agent, the exchange will be moved to the Conversations module.

9. The agent has an overview of the previous conversation the customer had with the bot and simply continues where the bot left off. If you had added visibility to agent on some of the created attibutes, those attributes will be displayed in Conversations.

10. From the customer perspective, the conversation continues seamlessly on their phone.

11. After the conversation is finished the customer leaves satisfied because the problem was easily solved.