Taking Flight with Chatbots


Find out how Lucy's experience with an airline's chatbot left a lot to be desired, and how testing could have prevented this.

I had an encounter with a bad bot the last time I went through an airport.

I managed to catch my first flight without any issues, but when I went to check in at the airport in London, I had a problem. The best explanation the person at the check-in desk could give me was that there was a glitch in the system and I’d booked a flight that didn't exist. The next available flight was not for another 20 hours. 

This was the moment when I realized I might be spending the night in the airport and the stress really started to kick in. To avoid that dreaded airport anxiety, it was important that I had my questions answered quickly. I logged into the customer service portal for the airline and as I began chatting with the customer service bot I was met with everyone’s worst chatbot nightmare: ‘I’m sorry I didn’t understand your request’.

To make this worse, the bot didn't seem to understand how to escalate my issue to a live agent. I don’t know how you’d react in this situation, but I decided there and then that it was the last time I was flying with that airline.

We’re all aware of the benefits of chatbots in reducing pressure on agents and ultimately saving costs, but what about improving customer experience (CX)? Airlines know that excellent CX fosters loyalty and they also know that all it takes to destroy that loyalty is one stressful experience. Chatbots are quickly becoming the first point of contact between customers and airlines; so bots need to be designed to make travel experiences - from booking to landing - smoother, more efficient, and ultimately, stress-free.

Now, back to my airport chatbot nightmare… I wondered if the airline was aware of all the stress it was causing? That there was a frustrated customer who wouldn’t be traveling with them again? Did they know how many other customers had a similar poor experience? And most importantly, did they know what they could do to improve and prevent this - bar removing the chatbot altogether?

➡️The first step is being aware of what chatbot technologies can actually do, and crucially how they can fail. 

➡️The next step is to systematically and continuously test the chatbot to ensure that it doesn’t fall foul of any of these common issues.


The potential and pitfalls of chatbots

Here are some of the areas where chatbots can improve CX for passengers, along with the associated pitfalls that may prevent them from achieving a positive CX:

🤖💁‍♂️Instant assistance: Chatbots provide instant assistance, available 24/7, allowing passengers to resolve their queries and concerns at any time. Whether they need to make flight changes, inquire about baggage allowances, or request special assistance, chatbots can provide prompt and accurate information, saving passengers time and minimizing stress. 

⚠️Chatbots that become overwhelmed too easily, respond too slowly or can’t handle large volumes of concurrent sessions can negatively impact this availability. To avoid this, airlines must validate and understand the concurrency level of the chatbot they are deploying - can it assist 10 customers at a time? 100? 1,000? More?


🎁🌟 Personalized recommendations: Chatbots can be equipped with artificial intelligence (AI) and machine learning (ML) capabilities, allowing them to learn from previous customer interactions and provide more personalized recommendations. From suggesting the best flight options based on preferences to recommending nearby attractions and activities at the customer’s destination of choice.
⚠️ However, there are instances when chatbots struggle to grasp the precise context of a conversation. Chatbot testing solutions are necessary to ensure accurate personalization and comprehension of each interaction. For example, if the bot recognizes that an individual traveled solo, it could suggest relevant individual activities instead of group ones. Chatbots commonly leverage a number of different data sources to acquire and make sense of this data, for example the airline's CRM (customer relationship management) database.


🛫📲 Real-time flight updates: I now know all too well that flight delays and cancellations are the unfortunate realities of air travel. Provided they are working correctly, chatbots can be used to keep passengers informed and updated in real-time. They can notify them of any changes in flight schedules, gate announcements, or even weather conditions at their destination. After all, keeping passengers well informed is half the battle when it comes to a good CX.

⚠️Yet, many chatbots continue to provide incorrect information - often very confidently! Airlines must list the topics they’d like their chatbot to handle, and make sure it gives the proper information back to the user in each case, based on the user’s respective information and current context (for example, in my case booking a non-existent flight and needing assistance).

✈️🔒Seamless booking and check-in: Chatbots can streamline the booking and check-in process, making it quicker and more convenient. Instead of navigating through complex websites or waiting in long lines, travelers can engage with chatbots to find and book flights, select seats, and complete the necessary check-in procedures effortlessly.

⚠️ Some chatbots may experience defects due to regression following the addition of new features or updates. Airlines must test and then test again with every update they make to their chatbot. They also need to validate that it works well in different scenarios and via the various channels they offer.


🗣️🌐 Language support: Traveling to a different country can be intimidating, especially if you don't speak the local language. Chatbots are able to bridge this communication gap by providing multilingual support.

⚠️Chatbots that only understand the local language are not much use to international travelers. If airlines serve customers outside of their locale, then they must make sure to have a multilingual chatbot (and obviously - test that it works as expected in each of the languages that it supports).


While each of these are excellent use cases for chatbots; for each there is also a corresponding pitfall - which highlights the key point to remember:


Chatbots are software, and software must be tested!

The only way for chatbots to be completely trusted by airline customers is if they provide accurate, timely 24/7 assistance; and this can only be achieved through rigorous AI chatbot testing, monitoring and training.

From providing instant assistance, personalized recommendations, real-time updates, seamless booking, and multilingual support, airlines are using chatbots to revolutionize the travel experience. They are quickly becoming indispensable tools for airlines in providing stress-free journeys for passengers around the globe. 

With chatbots taking care of routine tasks by providing valuable automation and assistance, and Cyara Botium looking after the chatbots, travelers can better relax and enjoy their journey.