While it’s never been more exciting to be in product tech, with new tech such as generative artificial intelligence to spice things up even more, it’s also getting more challenging to fit a growing “multiverse” of products and content into a world of shrinking screens.
Just ask Amy Wei, senior product director at Trip.com Group, who’s responsible for “public products” at Trip.com, whose Chinese counterpart Ctrip has been putting together the “connected trip” long before it entered the vocabulary of Western online travel agencies.
Trip.com has similarly been putting together a one-stop shopping platform to enable its customers to buy everything from flights, to hotels, transportation, including cars and rail, and tours, activities and attractions.
Add to that, an additional content layer as it builds an eco-system within the app to keep customers engaged from dreaming, to planning, booking, servicing and sharing during and post-trip, until they return for the next trip.
“My team designs the home page and order of product line and we had to do several redesigns to fit into the mobile screen, to reduce noise for users. It is getting more and more challenging,” she told WiT, adding, with a laugh, “Every product wants to appear on the first screen. We get questions like, ‘Can we get a pop-up here?’ or ‘Why can’t my product show up here?’”
Subscribe to our newsletter below
Yet at the same time, it’s never been more exciting.
Wei, whose team launched AI travel assistant, TripGenie, within the Trip.com app in February 2023, said, “Witnessing the integration of generative AI into our services has not only been exhilarating but has also opened vast avenues for product innovation. It’s indeed a golden age for product managers in the online travel industry, propelled by the rapid evolution of GenAI.”
Before we get into all that though, I was curious as to how she balances between customer experience and revenue optimization.
“We focus on the core business – flights, hotels, flight + hotel, trains – in some markets, car rental is important, like in Australia. In some markets, tickets to attractions are important. We also look at traffic and click-throughs and user intentions for product line, and metrics such as quality of product and price competitiveness.
“We need to design for user demand as well as for business interests. We get pulled in all directions, so we have to balance that.”
The content play – the second layer of content on the app – is integral to Trip.com’s mission of building and embedding a social network within its app. This layer includes Deals, Trip.Best, Travel Guides, Trip Moments and Trends – and it aims to guide customers from the top of the funnel down to bookings.
TripGenie – any AI tool can create an itinerary, it’s the quality that’s critical
And then comes TripGenie – in June this year, it announced three more languages – Italian, Dutch and Malay – to add to the list of English, Chinese, Cantonese, Japanese, Korean, Thai, French, Spanish and German.
It plans to eventually support all languages featured on Trip.com. The service will continue to expand, introducing capabilities such as real-time voice translation and enhancing pre-trip and post-trip services, such as easier hotel comparison, enhanced after-sales consultation and travel experience sharing.
According to Trip.com, users who use TripGenie to recommend destination and itinerary options are 30-40% more likely to return to the platform with twice the order conversion rate. These same users also spend approximately 20 minutes more on the app than the average users.
We are able to do more with less, but this product needs deep integration with Trip.com, so we need more hands on deck. I see GenAI as giving me a new arm; I am fascinated by the advancements in tech.
Amy Wei — Trip.com
Unlike other trip planners, Wei said Trip.com’s intention is “to keep all interactions seamless and integrated within our app, enhancing both coherence and convenience. We dedicated last summer to refining AI-generated itineraries. Today, any AI tool can create an itinerary. So the quality of the itinerary is the most important thing, and a tool to manage the itinerary properly is also a very practical need for travelers.
“This year, we’re doubling down on enhancing not just itineraries but also integrating other crucial services like flight and hotel bookings and post-booking support. Our use of advanced technologies like speech to text (STT), retrieval-augmented generation (RAG) for improving responses related to customer service knowledge base inquires and named entity recognition (NER) to deeply link to relevant properties, [points of interest], etc., ensures that TripGenie assists travelers efficiently without overwhelming them. The interface innovations such as the floating button and the 3/4 expandable overlay screen have also been pivotal in balancing user interaction and task efficiency.”
How TripGenie differentiates from other trip planners too, she said, is it integrates content from large language models (LLMs) and its own real-time user generated content including images and videos. “We build all the information, then use all the elements we have so that users do not need to jump from one website to another."
On lessons it’s learned since the launch, she said, “One of the challenges is people don’t know what questions to ask so we look at how to lower the barrier – help people ask questions. We do this with a welcome pop-up and when we see the customer has booked Singapore, we generate potential questions, pre-set questions based on user behavior and history.”
It also tries to find scenarios where users face difficulty and solves them. “For example, we see people spend a lot of time comparing hotels, so we launched Hotel Compare, and we use LLMs to compare location, facilities, based on #Kid-friendly, #Business travel, #Value for money or other preferences.”
Before TripGenie, there was TrainPal
Wei, who worked with Motorola and Nokia, and then Amazon for 1.5 years before joining Qunar in 2015, is no stranger to building verticals for the global market.
In February 2018, when she was CEO of International Train Ticketing at Ctrip, she launched TrainPal, a mobile app offering international train ticket booking services in the United Kingdom. It quickly expanded across Europe – Germany, France and Italy – and to Asia, South Korea, Japan and Taiwan.
“It is the second largest train platform app in [the] U.K., with 1 million downloads,” said Wei. In building TrainPal, she leveraged her experience in Qunar where she had learned to build a global distribution system middleware layer to connect airlines and distributors via API. She had also built the train product for the China market, when she was CEO from 2017 to 2021.
But building for China is different from building for international markets. “Trains are more complicated, they are also more domestic, and the system for each country is different. We needed to build a unified GDS with multiple languages and multiple payment options.”
TrainPal’s innovation was “Split Fares,” a unique algorithm that immediately calculates whether users can save money on train tickets by splitting the fare in two or more and then issuing two or more separate tickets.
The advantages of adopting AI and innovation in China
Wei is also used to building with lean teams. When building TrainPal, Wei started out with a team of four, which grew to 10. In building TripGenie, she has a team of seven. “We are able to do more with less, but this product needs deep integration with Trip.com, so we need more hands on deck. I see GenAI as giving me a new arm; I am fascinated by the advancements in tech.”
And if you think trying to keep up with generative AI advancements globally is tough, Wei also must keep up with advancements in China, which is taking place at an even more rapid pace.
“There are so many LLMs, everyone is building their own – Alibaba, TikTok, Tencent. There’s been a breakthrough in one of the new players, called Kimi, which started to introduce in October last year the ability to input 200,000 Chinese characters – one Chinese character is roughly equivalent to two tokens, English is one – so that has ramped up the competition. There’s a lot of diversity in innovation right now.”
Asked what advantages companies such as Trip.com have over other OTAs in embracing generative AI and other new tech, she said, “Flexibility – we embrace everything. TripGenie was launched last February, ahead of everyone. We’ve done so many new things in the past two decades – we have the mindset that everything is possible.”
She said there’s also a healthy supply of talent in product management and engineering. “We also have a good environment for innovation because users like to try new things, so it allows you to experiment and reiterate quickly. Plus, of course, it’s a big market.”
Her biggest concern is privacy. “This will depend on the discipline of the developers to use personal data responsibly.”
She added, “It comes down to, how do you want to treat your customers? Transparency is crucial for us. When we want to try to upsell something, we make it crystal clear. We do not try to mislead customers. It is really up to the discipline of each company.”