KLM and WestJet Deploy Chatbots To Enhance Customer Conversations

Written By: Jerry Witkowicz   jerry.witkowicz@180find.com

The rule based chatbots are now widely used by many companies across many business verticals. Conceptually they perform similar functions and the level of sophistication and complexities varies depending on the how these chatbots have been deployed, how they learn and how much training data is used to continuously train these chatbots.

In this tutorial, we examine two chatbots KLM’s BlueBot, WestJet’s Juliet a customer facing chatbots.

KLM's BlueBot and what it does?

According to Martine Van Der Lee, Social Media Director at KLM, since KLM added messaging platforms, KLM experienced double digit growth in customer conversations. This showed that KLM’s customers were shifting their conversations to private messaging channels. They were looking for personal conversations. KLM was experiencing a yearly audience of 140 million that included visits, message, customer questions, etc….

KLM recognized that majority of conversations about KLM on other platforms like Wechat,Facebook,Tripadvisor,Google, were owned and controlled by these external platforms and KLM concluded that conversations about KLM on other platforms should be owned by KLM.

KLM integrated its systems with these platforms to gain control and own conversations with its customers about KLM and give its customers personal service. As the volume of conversations continued to grow, KLM quickly recognized that it also needed to explore the use of smart technology like AI to automate some of the repetitive conversations and this was the beginning of BlueBot, KLM’s chatbot.

When BB was first deployed in 2017, it started with a simple rule-based system that could book flights for its customers. BB was initially trained with the company’s rich historical data that contained years of real text conversations with its customers and media mentions. This rich data enabled KLM to train BB to handle many real Questions & Answers and make flight bookings for its customers.

 

Source: Dialogflow

Since its inception, BB has evolved to become a smart assistant that uses Artificial Intelligence to self-learn by observing conversations and inputs from KLM’s live agents.

KLM’s BB is also integrated with Google Assistant which enables customers to connect to BB through Google assistant. In this function, customers can use voice commands to ask BB to find flights, book a ticket, receive help with packing their bags.

 

The KLM application on the Google Assistant

BlueBot Results

Today, the airline sends 65% of flight updates, 40% of boarding passes and 20% of booking confirmations via messaging apps. The self-learning system becomes 5% more accurate each week. Together with the tam of over 300 live agents, BlueBot is part of the KLM’s customer facing family of live and virtual agents.

Sources:  Martine van der Lee presentations

WestJet Juliet and what it does

In 2016 WestJet set up a dedicated team to monitor social media feeds and to respond to its guests and customers on Facebook and Twitter. This Social Care team comprised of WestJet team members with backgrounds from cabin crew to airport staff who could help its guests’ travel needs via the phone, email or other methods of contact so that they could answer questions about flight schedules, destinations, baggage info and more.

In August 2018, WestJew added its virtual assistant called Juliet designed to help the dedicated team serve its guests and customers. Using Facebook Messenger, guests and customers could interface with WestJet’s through the new chatbot called Juliet which is able to; provide travel inspiration and recommend destinations based on customers interests and vacation style. Juliet also helps guests explore destinations and when ready help book the flights.

Starting with simple tasks, Juliet continues to evolve and accumulates knowledge based on conversations she has with guests and learns from these conversations over time. WestJet has people monitoring Juliet and their agents intervene into a conversation if needed.

Juliet on Messenger Source: WestJet

In Nov,2019, WestJet announced Juliet, its digital travel assistant that was available on Google Assistant. This integration gave guests and customers a hands – and screen-free way to have a chat with Juliet and ask questions.

Some of the questions Juliet can answer on Google Assistant include:

“Hey Google, ask WestJet the flight status for WS123 today”

“Hey Google, ask WestJet about travelling with my dog”

“Hey Google, ask WestJet what identification I need to travel”

“Hey Google, ask WestJet about baggage costs”

Juliet Results

In an interview with Forbes, Alfredo C. Tan, Chief Digital & Innovation Officer at WestJet shares WestJets’ success story about Juliet.

“The launch of Juliet, WestJet’s newest Google Assistant, now has a front-end voice experience. Juliet’s core intelligence can now be exposed to other channels and interoperate with other applications like Alexa, Google Home, Instagram, WhatsApp, email and mobile etc. enabling a variety of inputs while collectively making Juliet perform better

It’s still early day and Tan acknowledges Juliet is very much still in the supervised learning stage. This means that if Juliet is unable to solve an issue, the session will be referred to a live agent to minimize guest frustration. Currently, 25% of guest queries are forwarded to a human agent. These instances are sent to WestJet’s partner Netomi to optimize the model.”

Sources:

Summary:

Number of key observations is derived from these two case studies. One is that each airline deployed its chatbot to perform simple selected task and evolved it over time. Second is that training data is key to deploying initial chatbot even to perform a single task. And 3d, chatbots do need to be monitored by human agents to help the chatbot learn and evolve.

Finally, in both case studies, chatbots continue to help each airline improve its service to their customers, and it’s clear that their customers are permanently shifting to digital channels.