We are all reading lots of column inches on how AI will impact our lives and much of the focus is around autonomous vehicles. Fortunately there hasn’t been that much on autonomous aircraft and, for now at least, I personally find that reassuring.
So whats the medium term impact of AI for the traveller? Certainly some of it will be around automated assistance – chatbots that handle simpler natural language based interactions with travellers at probably any stage of the journey. These already exist and can handle core traveller questions but they have an escalation route to their human counterpart.
Is that it? Well, the short answer is not by a long chalk. AI is often thought of as being about making smarter decisions and that may be true but I tend to take the view that it’s actually more about making good decisions a lot more quickly and most importantly in an automated manner.
So how does that impact the traveller? It actually means that any time a traveller asks a question from their travel company, that decision could now being taken by an AI driven system that is evaluating lots of data to make a much more personalised on-the-spot decision whereas previously the outcome might very well have been simply predetermined by some fixed rule.
Take ‘dynamic’ pricing of airline tickets as an example. The objective is to maximise the overall revenue for the flight. A classical approach might be to have a set of price buckets pre-configured so that as one ‘bucket’ of seats sells out the next higher priced bucket become available. I can easily picture the “only 2 seats left at this price” messaging. If the higher price bucket fails to sell and the flight departure date is getting close the pricing system may need to switch back to the lower priced bucket to encourage bookings. Revenue Management systems apply complex algorithms to advise Revenue Manager across thousands of products/buckets. Be in no doubt, this works but, if we are honest, this is quite clunky – it is effectively the same old seat-on-the-plane product at the best price the market will bear. The rest of the flight booking process is then followed up with the up-sell offer of ancillaries.
AI has the potential to change this quite substantially. By considering a much wider range of input data sources, much of it relating to the traveller themselves, an AI system can automate the customer-centric offer and pricing strategy. By that, I mean an AI based pricing system will be evaluating each flight search based on the surprisingly significant amount of information it has available at the time of the search. In addition to wider market and recent booking history date, there is lots of information about when the traveller is planning to depart, what day of the week it is, approximate time for departure, how long they are staying, is there a weekend involved, who are they are travelling with etc as well as how far ahead they are looking. All of this information can be captured and fed into a AI system so that it can make an automated decision about where the individual is in their decision making process along with what to offer and at what price. The number of potential product offers (including relevant ancillaries now bundled upfront) and price combinations will actually be very large but the AI system will make the best on-the-spot offer. The marketing blurb supplied with the now customised offer will also vary in its content and emphasis. This is all about the right product at the right time at the right price.
Sounds like a plan? I can absolutely believe that AI will be a key factor to the airline success not just in revenue but also in relevance and loyalty.
Are there any challenges? AI is only as good as the input data it has available. Capture and processing of the data is key. The good news is that IATA’s NDC messaging standard for airline searches/bookings has opened the door to not only accessing the relevant data for AI but also enabling truly on-the-spot dynamic product/pricing offers to be made.
I think we can all expect our travel planning and booking to be significantly impacted by AI and that is just in the short to medium term.
Triometric has recently launched Trio Data Engine, a real-time XML data processing component that can extract and process relevant NDC data at scale for feeding into a wide range of systems. That includes AI driven processes.