How are chatbots changing the conversation between marketers and consumers in air travel?

Chatbots. Since the surge of hype surrounding this AI-based technology in 2016, there have been untold amounts of speculation about its potential: for customer service, for customer acquisition, for personalization, for making sales and payments.

Chatbots aren’t suited to every industry, and in some they’ve quickly died out, leading commentators to dismiss them as nothing more than a brief fad. But there are other industries where chatbots look set to have a genuinely transformative effect. One of those is the travel industry, which combines elements of retail and customer service to create a bespoke experience for each consumer, and in which chatbots are beginning to take on a role that resembles the travel agents of the pre-internet era.

At Aviation Festival Europe, marketing executives from four different companies discussed the future of this ‘conversational commerce’ in the airline industry, and how it is changing the dynamic between airlines and passengers.

The participants included Jonathan Newman of Caravelo, which offers an air travel chatbot named Nina; Carolijn Hauwert of KLM, which has chatbots on Facebook Messenger and WeChat; Youvraj Seeam of Air Mauritius; and Guðmundur Guðnason of Icelandair, whose Facebook Messenger chatbot allows customers to search and book flights as well as discover interesting facts about Iceland.

The brands discussed their individual experiences with using chatbots, their benefits and drawbacks, the potential of the technology, and the best chatbots that they had personally encountered in the airline industry.

Uber for air travel

Using chatbots instead of a human customer service agent generally involves a trade-off between speed and empathy, observed Carolijn Hauwert: you can either opt for the speedy response of a bot, or the empathy of a human agent.

However, it’s also possible to have the best of both worlds by using a different agent for each part of the process. KLM noticed that because every customer service query began with gathering the same basic information, such as the customer’s booking reference, it was possible to have a bot manage the first part of the query, before forwarding it on to a human agent to manage the more complex steps.

Chatbots can also have some advantages over human agents when it comes to that initial response. Caravelo’s Jonathan Newman noted that Facebook Messenger chatbots have the ability to detect a user’s location and the language they are using, allowing the bot to immediately respond in the correct language.

To Newman, chatbots for travel have the potential to bring the same speed and convenience that Uber brought to the taxi industry, and that convenience is long overdue. “You can book and pay for an Uber within 30 seconds. Why can’t that be the same for air travel?”

What if you could book your air travel as easily as booking an Uber?

However, many in the industry need more convincing before they are won over by the potential of chatbots. This can be particularly true of top-level management, who are more cautious about buying into unproven technologies, and need hard proof of the benefits before they are willing to invest. In this situation, how can marketers go about convincing the top brass that chatbots (or other emerging technologies) are worth their while?

Youvraj Seeam advised that the best way to go about this is to use case studies from other brands. Marketers should take some use cases that are proven in other airlines, including the benefits that they are deriving from using a chatbot, and bring those to management. (An additional advantage of this is being able to learn from the mistakes that other brands have made, and avoid making them yourself!)

He recalled that when it came to proving the value of ecommerce to the brand, the e-sales team at Air Mauritius showed the brand management what contribution they had made to the brand’s overall revenue: as a percentage of brand revenue, e-sales had risen from 2% to 10% thanks to the team’s efforts.

But when it comes to proving the value of chatbots, it can be difficult to know which metrics to track. So how can brands quantify chatbot success?

Metrics for success

Chatbot technology is still in its infancy, and so even those brands who were early adopters of chatbots are still refining their approach.

Icelandair’s Guðmundur Guðnason reported that one year on, the brand is starting to see some results from its chatbot, but that they haven’t always had a lot of visibility into how well it’s working. Icelandair keeps track of how many bookings have been made, and tries to assess whether they are actually moving the load from the call centre onto the bot, which is a good indicator of how effective it’s being.

Carolijn Hauwert added that KLM has visibility over whether a response to a customer is sent by a bot or by a human agent, allowing them to see how many queries the bot is handling and at what point in the process a human agent needs to take over.

Jonathan Newman pointed out the difference between metrics and KPIs; while tracking how many queries a bot is responding to is all very well, is the bot delivering what the customer wants? How are they ultimately converting? How many steps do they wind up taking before they purchase? Ultimately, these questions can be more important in deciding whether chatbots are beneficial to a brand.

He noted that customers have a tendency to speak more freely when they know they’re conversing with a bot – which can be enormously beneficial to the brand, allowing them to gather a lot of useful information from what the customer says, and their tone.

This confirms the potential of chatbots as a means of gathering customer data, and could give marketers some additional evidence to bring to the table when trying to quantify their chatbot’s overall benefit to the brand.

Which are the best bots currently around?

Finally, the panel concluded by discussing the most effective chatbots that they have encountered to date. It was generally agreed that at this point in time, chatbots tend to work well for a very specific use case. For example, Lufthansa’s bot, which is called Mildred, can’t tell you whether or not you can bring your dog on a flight, but it can give you the best prices from London to New York.

Youvraj Seeam cited Aeroméxico’s chatbot, Aerobot – which was the first airline chatbot to be launched in the Americas – as “one of the best that he had encountered”. Aerobot, a Facebook Messenger chatbot which allows customers to search for flights in both Spanish and English, is another chatbot that has been designed and refined to carry out a very specific task, and does it very effectively.

Aeroméxico’s chatbot speaks both English and Spanish

Carolijn Hauwert said that as the industry has yet to produce a well-rounded chatbot, she doesn’t have a favorite. However, she had a piece of advice for brands who want to make a chatbot that excels.

Chatbots aren’t yet capable of being all things to all consumers, but as we have seen from the examples mentioned above, they don’t need to be. Brands can gain a reputation for chatbot excellence even if their chatbot is only capable of performing one or two tasks – as long as it performs them effectively.

The key is to make sure that you are setting expectations for your chatbot at the right level. As long as you don’t overpromise, you can dedicate your efforts to meeting those expectations – and wow customers by exceeding them.

Related reading

Big Data & Travel