Customer Service Chatbot Success Stories to Learn From

Chatbots for customer service are still new and unfamiliar to many, and it can be difficult to identify use cases that customer service chatbots could be useful for. So we’ve put together a profile of four very different customer service chatbots to showcase some examples of what successful chatbot customer service can look like, and what key lessons you can learn when formulating your own chatbot strategy.

Domino’s Dom for pizza ordering on any device

Domino’s was the first major pizza chain to implement digital ordering via chatbot – ahead of both Papa John’s and Pizza Hut. The Domino’s bot is named Dom, and is accessible through a wide variety of channels, including:

  • Slack
  • Facebook Messenger
  • SMS
  • Apple Watch
  • The Company’s mobile app
  • On Twitter, by tweeting the [pizza] emoji
  • Google Assistant & Alexa
  • Smart TVs

Some channels, like ordering through smart watches or Twitter, rely on setting up an Easy Order profile through the Domino’s website. However, once the Easy Order profile is set up, customers can use Easy Order on any of the channels it is available on and earn rewards points for free pizzas after a number of purchases. Ordering is also ridiculously easy, and fun. Customers can order by sending the 🍕 emoji either over Twitter or to the Domino’s SMS shortcode to order their pre-configured Easy Order. The bot will respond with price and ready time, and customers simply need to respond within 20 minutes with either “confirm” or the 👍 emoji to finalize their order to have order status and ready time confirmed.

Additionally, prior to the 2017 Super Bowl, Domino’s added full-menu functionality to their Facebook Messenger bot – for a menu-guided tour through the full Domino’s menu, allowing for a fully customized order, rather than being restricted to a pre-configured Easy Order. Customers can use the Facebook Messenger bot to order pizza, wings, and other menu items, and customize their order.

Perhaps the only drawback to using Dom for pizza ordering is that the ability to pay electronically at the time of ordering is very limited. The only channel that enables digital payment at the time of order is Apple Watch. (If customers ordering on another channel want to pay digitally, they must pick up in-store.) With the Apple Watch app, you can tap the Domino’s icon to place your pre-configured Easy Order and pay using Apple Pay. Once payment has been confirmed, your watch will display an order tracker showing the status of your order.

Key Lesson: Make it ridiculously easy for customers to perform key interactions on any channel.

CNN’s content bot, providing personalized news

CNN first launched it’s news content-suggestion bot in 2016, largely because they didn’t want to be beholden to Twitter and Facebook for the discoverability of their content, since both platforms change their content algorithms on a daily basis. The CNN chatbot delivers content recommendations based on user preferences and interest from past activities. You can also message the bot to access news stories about a specific topic. Over time, the CNN chatbot will learn your preferences and be able to deliver stories more tailored toward your personal interests.

Alex Wellen, CNN’s senior vice president and chief product officer, says that the goal for their chatbot was to create a way to deliver real-time, personalized news. “We want the conversation with CNN to feel personal and non-intrusive. That means finding the right balance of notifications and giving the audience control of how often or how few alerts they receive each day. In other words, the chatbot can’t be too chatty.”

Like Dom, the CNN chatbot is available on a variety of messaging apps like KiK, LINE, and Facebook Messenger. A version of the bot is also available for Amazon Alexa, where people can ask open-ended questions about the news. An important key to CNN’s chatbot success across channels, though, has been CNN’s approach to tailoring the bot for different platforms. Says Wellen, “each CNN chatbot has its own personality and functionality, in line with what we know about the audience and that audience’s typical behavior.”

Being willing to monitor and adjust bot output to meet the social conventions of each social platform has delivered real benefits. And interestingly, the usage results have varied by platform. The bot has seen the most growth on LINE, but has the highest engagement on KiK. Additionally, the most experimentation with the bot’s capabilities happens on Facebook Messenger and Amazon Alexa – which both allow the user to ask open-ended questions about the news.

Key Lesson: Pay attention to how customer preferences vary by channel, and adjust bot output in response to complaints

Westjet’s Margot, a digital travel assistant

The popular discount Canadian airline Westjet was already a leader in digital customer care – it was the first Canadian airline to provide customer support 24-hour customer support 365 days per year over social media. But, not content to rest on its laurels, Westjet hired Alfredo Tan at the beginning of 2018 as the new Chief Digital and Innovation Officer. Since then, Tan has pushed the company in new and innovative directions by using things like hackathons to identify digital solutions for CX pain points.

One of Westjet’s new digital initiatives is the launch of Juliet – a Westjet digital travel assistant named after the first aircraft owned by Westjet. Newly launched at the end of August 2018, Juliet can be accessed through Facebook Messenger and provides day-of information for booked flights and can give instant answers to questions about things like baggage guidelines, flight status, and traveling with pets. Juliet can also assist booking flights, and can help users find inspiration for travel destinations by answering questions about their ideal trip. Westjet also plans to expand Juliet’s functionality over time.

Juliet learns from interactions with guests, and is designed to become more knowledgeable over time. However, a key part of Westjet’s implementation strategy is proactively setting customer expectations. In announcing the new chatbot (link: https://blog.westjet.com/introducing-juliet-our-new-chatbot-assistant/), Westjet laid out the use cases and benefits, but they also reminded their customers that Juliet is still learning and their human agents aren’t going anywhere:

Our Social Care team is ready to assist

We expect that Juliet may need a little help along the way. That’s to be expected from a brand new chatbot, who is still learning the ins and outs of airline social media conversations. That’s why our 24/7 Social Care team is on hand to pick up the conversation should Juliet not quite understand what you’re asking, or if she simply doesn’t know the answer yet.

Westjet’s announcement is a great example of setting expectations! Chatbot technology delivers real benefits for CX, but machine-learning is far from a perfected technology and any bot in the new stages of implementation is bound to make some mistakes. Additionally, most consumers are still unfamiliar with customer service chatbots and might not be aware of what sorts of interaction problems they should expect – so assuring consumers that humans are always available to help is definitely a good move.

Key Lessons: 

  • Start with easy-to-tackle use cases, then expand to more complicated use cases. 
  • Set expectations, and make it easy for customers to switch to a human agent.

Lidl’s Margot – wine recommendations with fun personality

Also new in 2018 is the wine recommendation chatbot launched by UK grocery store chain Lidl. In February, Lidl announced the launch of their bot Margot (a play on the common name as well as the Margaux wine-growing region in France), that will help customers select the best wine available in-store based on either their budget or the ingredients of a meal they are planning, for instance – “which red wines from Chile under £6 do you sell?” or “what goes with grilled salmon?”. The bots capabilities are surprisingly robust. It can recommend 220 individual food/wine pairings, and can answer questions about 640 unique grape types.

The chatbot is conversational, and uses a mix of natural language processing and menu-based flows. The conversational style posed some interesting development challenges, given the bot’s ability to filter recommendations based on price. For instance, it had to be able to recognize intents for all of the following: “£5”, “5£”, “5 quid”, “a fiver”, or even “around five-ish”.

And while it’s more of a fun feature than a key function, Margot also can make food/wine pairing suggestions based on emoji. Margot recognizes 88 different emojis; users can ask for the best pairing with 🐔 or 🍕 and Margot will respond accordingly.

Key to Margot’s success has been her high discoverability. Margot is linked to from the company’s site, which provides clear instructions on how to launch the bot using Facebook Messenger. Additionally, Margot isn’t left to converse with customers unattended; human agents monitor bot interactions for requests to speak with a human, and customers always have the option to speak with live staff with their wine questions.

Key Lessons: 

  • Make the bot easily discoverable.
  • Monitor bot interactions for customers who want to speak with a human agent. 
  • Don’t be afraid to build with whimsy.

How can you get started with your own customer service chatbot?

Stay tuned for our next post, in which we’ll have buzzword-free advice and best practices for getting started with building your own chatbots for digital customer service. In the mean time, if you’d like to speak to one of our digital customer service experts, you can contact us here or ask to book a demo.

You May Also Like

About the Author: Anna Kreider