AI “most powerful tool” – 15% efficiency improvement in engineering, 80% of queries answered by bots, 95% answered within 20 seconds

“Use tech to solve the problems and commercial will follow”. That was the central message delivered by Xing Xiong, COO of Trip.com Group, to the more than 170 partners gathered at the group’s Airline Global Conference in Shanghai last week.

Calling himself a technologist at heart, Xiong said that despite the industry back to resuming the high growth it enjoyed pre-Covid, it was still facing many challenges and that “it is surprising to see why there are still so many problems out there, despite so many pioneers working on them for so long”.

Xing Xiong: “Commercial shouldn’t be a blocker to value creation.”

His message was, whether you are an airline, OTA or airport, it was in everyone’s interests to create value for the eco-system and for the traveller. “Commercial shouldn’t be a blocker to value creation,” he said. “In my career, commercial has never been a challenge.”

He then went on to list these issues – LCC tech challenges, NDC, Flight AI/Flight Information, Corporate Travel and AI Inside Trip.

LCCs – OTA share will rise, virtual interlining needs to be solved

LCCs now fly half the world, he said. In APAC, they account for 56% of capacity; in Europe, 52%. LCCs are more about direct sales, and he estimated direct sales at 60% with OTAs at 15%. “Many low cost airlines were against retailing through OTAs, but they don’t have to. Some have started changing, there are still stubborn ones,” he said.

“OTA share will trend up over time because LCCs are expanding into longhaul and cross-continents where OTAs have the edge. OTAs are consolidating, and they have the edge in efficiency and tech capability and the cooperation between OTAs and LCCs will become more constructive over time,” he predicted.

LCCs face retailing challenges in interline from the lack of infrastructure support to virtual interlining which he estimates accounts for 20% of traffic. He credited Kiwi.com for the good work it’s done in this area “but not the good experience”.

“Lots of work still needs to be done, lack of connection convenience, luggage check through, flight change protection. Look to book is high, building and running flights search is expensive, fare accuracy and product diversity are the sacrifice.”

Forseeing these challenges, Trip.com acquired Travelfusion in 2014. “We were only addressing problems on a point-to-point basis, that’s 60-70% of the problem. L2B, accuracy and product completeness issues are still challenging. On standards, there are many questions, not many answers.”

This is why Trip.com is exploring a partnership with Navitaire to explore commercial solutions. “As a technologist, that’s how we innovate. We don’t have to start with standards. We want to work with Navitaire to find a win-win solution. It’s very early but we are excited to work with pioneers.”

Elaborating on the low cost focus, Yudong Tan, CEO, Air Ticketing, Trip.com Group, said, “We believe there is a huge opportunity if we can solve distribution problems for low cost airlines, just as what we have done with Amadeus and other GDSes for full service carriers. The whole flight process is not only about booking, shopping is also very important.

“Airlines cannot really solve the shopping experience themselves, we need better solutions and OTAs can add value to LCCs.”

Tan also spoke about Trip’s virtual interlining solution, which combines content from multiple source, including airlines, GDSs, aggregators and consolidators, using its own algorithms to create new travel options beyond traditional airline interlining protocols.

Bryan Koh, Singapore Airlines: “NDC allows us to democratise our content so that they work as if they were another replica of my B2C channel.”

NDC: Bottlenecks in implementation

It’s evident that NDC implementation and adoption has been challenging, to say the least. American Airlines has had to dial back its aggressive policies, with chief commercial officer Vasu Raja departing and CEO Robert Isom saying other commercial changes are afoot as the airline looks to revise its distribution strategy.

Finnair, which declared in 2021 that it would no longer support EDIFACT-based distribution by the end of 2025, has announced a delay in arrival at the desired destination. “We continue this path, but the pace of change towards modern retailing in the airline industry has been slower than we originally expected,” Finnair acknowledged. “At this stage, we do not see it feasible to end EDIFACT by end of 2025 across all of our markets.”

For the first quarter of this year, Finnair reported that 69% of its passengers were handled via direct channels or through NDC-based connections.

At the panel on “Next Gen Retailing”, Bryan Koh, Divisional VP Ecommerce and Distribution for Singapore Airlines, supported Xiong’s call to “solve for tech first and commercial will follow”. “That’s the most important message,” he said.

“NDC allows us to democratise our content so that they work as if they were another replica of my B2C channel,” he said, adding that 40% of SIA’s bookings now come through NDC channels.

In his speech, Xiong said, “It’s all about cost efficiency for the whole ecosystem. The most efficient solutions will win, or should win. Commercial is just a piece of paper between ourselves, don’t let it distort value discussions. There is an overwhelming L2B challenge – booking is golden, searches is just cost. Operational and service expectations are not met, there is lack of economy of scale.”

Flight AI Data Product: Customers want more comprehensive flight information

He said that Trip.com had accumulated 100+ PB of data at a global level covering comprehensive products information from all lines of business and their history; user profiles, preferences and behaviours, orders; and UGC, POIs, tourism meta, data from partners, etc. “We have more than 800 data experts mining our rich set of data, empowered by more and more powerful technologies.”

He quipped, “Basically, we did not sleep during the pandemic, we invested in data tech.”

Its data platform for Aviation Market Insights is now used by 3,500 registered users, and 150 airlines and airports. “We created this for value, not for revenues. Some of our partners say they want to pay for this. They say, if we don’t pay, we don’t feel good,” he said, laughing. “But we do this not for profit.”

It also has a refined data analysis dashboard for diverse business scenarios as well as an automated tool/solution for revenue management.

He said there is increasing customer demand for detailed flight information. “They want precise and comprehensive flight updates beyond just flight times, accurate arrival timing predictions and real-time tracking tech that enhances passenger planning and satisfaction as well as comprehensive flight details such as boarding/parking gates, luggage belt, seat details and wifi.”

He called out a Chinese startup called VariFlight that does real-time aircraft tracking. It covers more than 800 airlines and 9000 airports worldwide and monitors over 10b flight trajectory data daily, covering the dynamic trajections of 66,000 aircraft worldwide. It now covers 85% of China’s aviation data application market, 60% of domestic airports and 80% of major domestic airlines. And it is expanding globally.

“It is digitizing the industry,” said Xiong.

Corporate Travel: Held Back By Complexities But There’s Big Demand

Xiong said that thus far, the success of TMCs has been limited by these challenges – tech challenges to fulfil very segmented requirements, supply system challenges that are difficult and costly, weak negotiation power, low customer loyalty and low margin.

“But there is very big demand,” he said, calling out Navan (formerly TripActions) as an example of a company trying to solve the problem. “This industry needs tech innovation. If you can solve it, people will love you. It’s a hard problem and it deserves our attention.”

He said Trip.biz was the group’s corporate travel bet. “It was a bold investment, we put more than 800 engineers on it, and we leveraged our competitive supply system. We are dominant in domestic but globalization is an essential part and we are still in an early stage.”

He said thus far, Trip.Biz has provided travel services to 15,000+ large corporations, 980,000 SMEs and over 52m employees worldwide. It is expanding to Hong Kong, Macau, Japan, Singapore and Vietnam.

AI Inside Trip: Staff count down, order volume up – 80% of issues resolved by chatbot

Calling OpenAI an “eye-opening data technology breakthrough”, he made the observation that it was legacy wisdom, combined with recent unprecedented computing power and that its statistics-based basics could limit its innovation capability.

But “so far, it is the most powerful tool, which has already brought changes unthought of.”

Trip is seeing software development efficiency improvement – code generation, unit-testing, code reviews – of 15% and in inspiration and innovation, it is experimenting with TripGenie on the demand side and exploring trip planning capabilities.

It is in customer service that it is seeing real results. In 2023, staff count was lowered by 17%, but order volume went up 46%. In 2024, staff count went up 8% while order volume grew another 26%.

“We are seeing higher efficiency and continuing improvement,” said Xiong. “Eighty percent of customer issues are fulfilled by self-service autoboots. The remaining complicated issues require deeper expertise and effort from our service staff.”

In total, the group has about 10,000 customer service staff working in seven customer service centres worldwide to handle more than 15m monthly omni-channel service requests. About 75% of cases are solved by AI, automation service coverage is at 97% and 95% of enquiries are answered within 20 seconds.

Yudong Tan on customer service:  “Still we need a human to get involved not only to just solve the problem but also they need the EQ, to handle our customers who find themselves in trouble.”

Yudong Tan does not see it ever getting to 100%. “Still we need a human to get involved not only to just solve the problem but also they need the EQ, to handle our customers who find themselves in trouble.

“We need to keep training our customer service people. We need to equip them with more efficient tools, more knowledge base. We also need to recruit higher quality people and so we cannot use average standards for recruitment anymore.”

Another area AI is being used is in AI-assisted translation where it has developed a Trip LLM where it is able to apply available high-quality documents for LLM training, use translation content in tuning for better accuracy and establish a multi-language tourism industry terminology database for continuous training and tuning.

Xiong inferred this leap in translation tech will give Trip a leg up in its globalization quest. He observed that while tech can be global, humans are not as fast to make that leap and given its engineering teams are based in China, this advancement in translation technology is significant in sharpening Trip’s competitive advantage.