Ai Editorial: Recommending it right – can travel triumph in a battle of its own?

First published on 17th March, 2017

Ai Editorial: Airlines need to find ways to make luring recommendations despite the inherent complexities associated with a category like travel. Time for airlines to look beyond their role as a delivery service from point to point, writes Ai’s Ritesh Gupta


The way we shop for non-travel products, say a book or a smartphone, invariably gets compared with shopping for an air ticket, hotel bookings, in-destination purchases etc.

The likes of Amazon and Netflix continue to be a benchmark for personalised recommendations for a vastly different sector like travel.

Is this comparison between retail and travel apt?

“Pure collaborative filtering made famous by Amazon will not work for travel products,” says a source. “For us, contextual cross-selling has been a key driver for ancillary product attachment for years.”

“Travel is about exploration, vastly different from retailing. Each of us can breathe travel in a peculiar way, probably unlike any other product. So recommendations can be refined, tailored to an extent, but can it be cracked 100%, a complex puzzle yet to be solved,” said another source, referring to the possibility of coming up with an equivalent of Amazon or Netflix in travel. “Of course, unlike aggregators, airlines, at least for air-related travel don’t offer options (flights from another airline) as of now. But certain carriers may broaden their product portfolio in the future!”  

Still, to an extent, comparison is justified as consumer wouldn’t differentiate experiences. Choices, convenience…all that makes shopping a breeze, consumers would expect that to happen with every transaction, be it as a retail shopper or a flyer.

Here we assess how airlines need to refine their own journey of coming up with apt recommendations:

·          Matching expectations arising from non-travel shopping environment: When Amazon, Netflix etc. come up with a recommendation (and that too, on a regular basis) that resonates with us; it builds affiliation with the interface/ app as well as the brand. In fact, the association with these companies (a majority of these not even own the products they sell) is strong owing to the choices they offer us, the way they are offered, the way our activity is remembered, frictionless transactions etc. The Amazon app or Uber app always remains at the top of consideration set.

Of course, travel as a product category is different. But it does put the onus on travel technology, product as well as marketing professionals to come together to refine and deliver a relevant recommendation to brighten up the chances of a conversion. Attaining such level of expertise would mean acting on data – own data, 3rd party data/ external data such as weather, social, media consumption etc. Not only this integration is must, but also what is done in terms of predictive analytics, pattern recognition etc. needs to be relevant. Hopefully the emerging crop of travel flight search start-ups would lend a new dimension. The idea of finding your next destination or journey is luring, but when I see a mail or notification enticing me to take a trip to a destination  that I neither searched for nor matches my “activity” it is neglected straightaway.

·          Being where a recommendation is valuable: How to be a part of a digital environment, be it for an ecosystem (say Tencent or Google) or on a specific messaging/ social networking platform, where users spend their time, their decisions get influenced etc. is what airlines have to capitalize on. Traffic generation has never been a forte of airlines, so they need to be proactive enough to make a recommendation where audience is. For instance, if a couple is interacting via Facebook Messenger or Skype or Whatsapp, and are zeroing on a destination, can they book their air ticket or accommodation within that interface? Can a relevant recommendation from airlines at an opportune time facilitate a transaction?

Airlines definitely need to come up with a robust API strategy, rather than letting the likes of travel meta-search engines to capitalize on traffic resulting from Facebook, Alibaba, Tencent, Apple etc. via integrations and “connectivity” to pave way for recommendations of the product actually owned by them. (Meta-search engines have been moving even further up the funnel via native app features such as traveller inspiration timeline and push notification of travel ideas. They even provide APIs in addition to widgets and white labels to power travel search). So airlines, too, can take such collaborative approach via APIs and work with custodians of traffic. What if a meta-search engine provides FFP of a traveller with a particular airline, can this airline come up with a real-time recommendation at the time when this traveller is on the meta-search platform?

·          Recommendations that aren’t about past but “now” too: Using semantic analysis and pattern recognition for driving repeat buying, or automating cross-selling isn’t new. But there is scope for improvement in areas like quality of recommendations provided data ingestion layer is robust. Airlines need to focus a meticulous data management plan i. e. collect, store, process, analyze, and visualize big data on cloud, and look at data that tells about “now” or the moment in context. In fact, the current shopping cart and the path the customer took to get there can tell you far more about what they do next than any number of prior bookings. With such plan in place, airlines can look at crafting a recommendation algorithm, encompassing various stages of travellers’ journey including real-time travel disruption management. Be it for accurately predicting hidden interests, evaluating minute behavioral changes or working out recommendations for various contexts, airlines have to capitalize on artificial intelligence and machine learning.  Quality of recommendations have gone up, too, avoiding erroneous attribution and refraining from displaying recommendations repeatedly.  

·          Smart recommendations with no heavy “background” analysis: Detailed profiling is without doubt a weapon in the arsenal, but what if there is some quick, light analysis “on-the-fly” that can delight a traveller. Say a traveller is watching a movie in-flight, how about coming up with a notification for buying merchandize, tours and activity-related buy or even exploring a new destination? Marketing technology is advancing at a rapid pace, and new possibilities are always exciting. Today an airline can bank on a recommendation emanating from a location recommendation engine, which is backed by a matching algorithm built upon geo-data-mining and machine learning processes. So say one arrives on, is asked few questions about interests/ preferences for that particular trip, and is recommended few destination options with authentic content and itinerary. Such location matching in the form a viable recommendation opens up a plethora of option, including monetisation via ancillary offerings.


Gain an insight into intriguing issues at Ai’s 11th edition of Ancillary Merchandising Conference in Spain this year. 

Date: 25 Apr 2017 - 27 Apr 2017; Location: Mallorca, Spain 

For more info, click here


Follow Ai on Twitter: @Ai_Connects_Us


  • Ai Video: Counting on people and agile adaptive digital platforms +

    First Published on 22nd May, 2018 Full service carriers have mainly focused on safety and operations for years. But it is time for them to be a part of a Read More
  • Ai Editorial: 5 areas that can propel airlines’ digital transformation journey +

    First Published on 14th May, 2018 Ai Editorial: Avoiding quick wins, adopting fail-fast attitude, opting for growth culture, a governance team and setting up a “best of breed” platform for Read More
  • Ai Video: Leveraging XML data for 3rd party distribution +

    First Published on 9th May, 2018   Airlines are on look-out for ways to evaluate search and booking behavior on every channel they are associated with. A critical aspect of Read More
  • 1
  • 2
  • 3
  • 4
  • 5