No longer is “digital transformation” a sufficiently catchy buzzword for airlines to convince the world they’re moving into modern operations. Today the play is in “artificial intelligence” solutions, with plenty of suppliers keen to help airlines realize their aspirations to be more like Amazon, Uber, or some other household name brand. And there are plenty of suppliers waiting in the wings to help with that transformation process.
Alas, even as airlines invest in these systems, they continue to demonstrate the troubles they have with implementing seemingly simple intelligence for their offerings. Perhaps the AI solutions will improve the situation, but it seems many of them could be addressed far better by simplification of processes and policies. Or leveraging existing logic rather than inventing new systems.
A few examples
Airlines want consumers to believe that having more information about the traveler will enable better targeting of offers. This was key to the NDC pitch for fare distribution – if we know you have elite status we can bundle bag fees in the fare! – a decade ago, and continues as part of the ancillary revenue push today. But is it working?
Take, for example, the opportunity to up-sell passengers during the booking process. Fare bundles enable that. But if an airline knows the passenger and knows that passenger already holds elite status, thereby getting all the upsell benefits for free, targeting them with that pitch is a waste. Or, even worse, triggers concern from the consumer that their benefits might have changed.
Upsell opportunities exist during the check-in process as well. Selling a passenger lounge access on a long journey seems a simple win. Pitching it (ore premium seating) to a passenger traveling in a premium cabin and with elite status – each of which already grant that same access being sold – is an unfortunate miss.
In this case it is not even that the airline does not know how to quickly determine if the traveler has access. The computers at the lounge can quickly decide based on a scan of the boarding pass. But without “AI” it seems this cannot be extended to the check-in flow.
Once on board, the opportunities should be even more abundant. Airlines know absolutely everything about a flight and the available services on board. At least in theory. So sharing a QR code on the IFE for passengers to follow should deliver details about that flight, right? If only.
Instead of flight-specific details, the link goes to a generic “about our offerings” page, leaving passengers to try to sort out which of the many categories their flight falls in to.
How the airlines want to use AI
Perhaps the above examples are unfair. Sure, they appear simple on the surface. Reconciling the passenger details, flight details, and available amenities does not seem like it should require advanced compute capabilities. So maybe the airlines have a better plan for what their real AI implementations will include?
Speaking at an investor conference recently Delta CEO Ed Bastian suggested the company will use AI to help reservations agents parse customer queries, reducing the amount of time it takes to answer a question such as policies for travel with a pet from five minutes to five seconds.
Martin Gauss, President and CEO at airBaltic sees similar value as the company deploys its home-grown and third-party developed AI solutions, citing an “ongoing aim to increase efficiency across different business functions.” To date, the company says the main usage covers “several question-and-answer type tools” which support new hire onboarding, and quickly finding information related to operations or policies.
An adjacent version of this concept is already in service, helping Hotel Planner reduce call times and, more importantly, reduce training required for agents before they’re able to handle live customers. CEO Tim Hentschel described it during the World Aviation Festival in Lisbon last November as like an extra agent also listening in on the call and surfacing links to likely useful policies on agents’ screens. It is not, at least for now, providing answers directly to customers, avoiding the risk of hallucinations or user-manipulated results in AI-powered chat bot solutions.
In both these cases the time savings should be welcome by everyone involved. Whether it is AI or speech recognition and picking keywords from a conversation and then calling that AI is perhaps a moot point to debate.
Moreover, especially for policy and fee questions (including the lounge ancillary noted above), airlines should already know the answer or have a system to quickly determine it. If Delta knows how to quickly calculate the charge for a pet at the check-in counter, shouldn’t that same data be available to reservations agents and passengers, without the added layer of AI? Ditto for automating the opening of the lounge gate when a boarding pass is scanned.
Pricing and ancillaries
Airlines also talk about using the new “AI” tools to help with pricing decisions. Whether that’s artificial intelligence or just more aggressive statistical analysis thanks to ever increasing compute capabilities is a conversation the AI proponents likely would prefer to avoid.
This sort of analysis is, perhaps, the area where such tools are more likely to be of value. But the idea is not new. Airlines have been using extensive data analytics to optimize pricing for decades, with each new program iteration celebrated as a watershed moment for yields. The main difference appears to be calling it AI now.
In some cases, this sort of shift is obvious, and potentially challenging for airlines. Amadeus recently launched Nevio, with Saudia and Finnair signed on as early customers. The platform bills itself as “Built on modular, open and AI technology with a new value-driven mindset” to provide “end-to-end solutions, beyond Offer and Order.” This transition includes pulling customers from Amadeus’s Altéa platform, which is not particularly old by airline standards.
Such migrations are not trivial, nor are they inexpensive to implement. And while the providers can offer optimistic forecasts of improved revenues, actually delivering that is far from guaranteed. Will the system deliver the correct ancillary bundles? Or will it continue to offer services irrelevant to passengers? That’s not a function AI can solve.
Similarly, will customers shift their buying patterns more towards airline-managed bundles rather than a la carte pricing of trip components? Even Allegiant, notoriously successful at selling air/hotel/car packages, recently admitted it has shifted its marketing for hotel bundles as consumers become more proficient in finding their own deals.
Improved data analytics and machine learning, couched under the hype phrase “Artificial Intelligence” offers potential for airlines to improve revenue. And maybe even pitch ancillaries to consumers in a smarter, more targeted manner. Historically, adding another layer of technology on to the pile of siloed data has not proven sufficient to smooth those bumps.
Is “AI” going to finally solve that? Probably not, but at least we’ll see some additional investments made to find out.
A favor to ask while you're here...
Did you enjoy the content? Or learn something useful? Or generally just think this is the type of story you'd like to see more of? Consider supporting the site through a donation (any amount helps). It helps keep me independent and avoiding the credit card schlock.
Leave a Reply