In-flight analytics - can I have my favorite chocolate mousse on-board?
This case study is about overcoming the disconnect between what travellers expect and the effort behind data, analytics and personalisation. Ai’s Ritesh Gupta explores the same with his recent flying experience
I love chocolate mousse. In the recent years, as an economy class passenger, when I have travelled to the U. S, be it for via Air France KLM or British Airways, I have had mousse as part of the in-flight meal. Taking two or even three back-to-back flights over a period of 24-30 hours is no fun. But whenever I had mousse it did refresh me, helping me remaining afloat for few hours at least. The hiccup here is that I am not too sure on which sector I would be served my favourite dessert.
And the worst part– the airline crew having it and not knowing I want it.
Moment of magic @ in-flight: I recently flew four long-haul flights in one trip via British Airways.
Outward: New Delhi to London to Fort Worth.
Inward: San Diego to London to New Delhi.
The chocolate mousse was served as part of my in-flight meal on the following sectors: London To Forth Worth and London to New Delhi sectors. The brand was same. On my last flight, I asked the air-hostess about the brand of the particular chocolate mousse. She didn’t know it, but she served me one more. It delighted me no end.
One way to feel happy when you are flying two back-to-back nine-hour flights.
So before we hear from experts about what can pave way for this not just being a co-incidence, let me share relevant details:
Trip planning, booking and journey
Booking was done by my company, Ai, via BA website. As for indications that I wanted chocolate mousse, I searched for all the sectors on Google and typed “Chocolate Mousse on BA flights”. This was done via my PC at home and my iPhone. This was done over a period of 2 weeks before my first flight. Other touch-points: I used web check-in for all flights via my PC at home and laptop while on the road, and I also reserved my seat (aisle). I also accessed my in-flight email (Gmail), scanned check-in image via my iPhone for all the trips.
Profile: I am a laid-back traveller, and I don’t think of accruing points/ miles for any loyalty program. As for BA, I travelled after eight years (email id was same).
Points to consider:
The context may be derived from external factors - such as looking for sunny destinations in Spain when it’s raining in Dublin, or internal factors - such as searching is generally done on a large screen when time to concentrate is available but firm decisions are made concrete with family members in the evening in a casual manner perhaps sitting on a couch with a mobile tablet device!
These are all things that can be correlated from analytics and understanding the internal and external factors that drive behaviour makes irrational behaviour predictable.
Rachel Besant, marketing manager, 15below (the company focuses on data sources which the airline subscribes to - i.e. the FFP, reservations systems, other third party providers including merchandising etc.) says if BA has a system which logs the food options available on their flights, then it is possible to integrate with this data source (whichever system - third party or internal solution the airline chooses) and offer/ pre-order chocolate mousse, should BA intend to offer this service. Airlines will be asking if this is a significant enough differentiator for the additional logistical complexity this adds.
How would it work?
15below, for example as a notifications and workflow provider, could directly hook into this data source and proactively send notification to traveller with option to pre-order their meal. They would then record this action back into the PNR (and other system such as CRM, depending on how BA want this to work). The tailored workflows would continue to record and extract traveller preferences in order to send very targeted, personalised and timely offers.
The actual decision to charge or to provide as complimentary should be based on a couple of items bubbling up from the analytics - loyalty status and potential lifetime value to the brand, what’s your recency frequency metric, your average order value, your service recovery score (have the airline messed up with you previously – e. g lost bag, cancelled flight, spilt coffee) , and your social influence score (how likely you are to tweet (or write an article) about the experience?
“So the analytics should be able to predict you want a mousse on the next flight - the decision or option to charge you based on your profile - inform the flight attendant that you are important to the brand and not to charge you for the pleasure,” shared O Brien.
Another executive told me: Realistically it probably wouldn’t be feasible to do any of this without pre-booking your meal (beyond just selecting ‘VGML’, for example). In-flight staff don’t have the ability to record what meal you selected, the source mentioned.
There are airlines that allow one to add chocolates / champagne for any wedding couples on their honeymoon as a pre-order component, to get a fillet steak in economy on transatlantic flights. Also, as per the feedback that I gathered, say if you’re sitting in economy but you’d like a First Class meal, you can now order this (at a cost). A specialist then arranges all the logistics and supply of these meals ready for the flight take-off. This is just one way for airlines to offer a more personalised service whilst making some additional revenues.
The key restriction right now is:
The future of in-flight analytics should be able to offer you the rest of the movie you watched (and didn’t finish) on the last flight - recommendations of movies based on what I have watched - this could be applied to food, drink, duty free products, gifts and home delivery products.
This is the same outside the cabin - offer me a specific coffee as I am entering the airport to have it available in lounge - (Air NZ are already doing this with a Barista type service), baggage location, transfers if I like it with an understanding of my destination (such as office or home locations).
As O Brien says, analytics will re-define the shopping and travel experience over the coming years and make it frictionless to offer and consume preferred products.
Follow Ai on Twitter: @Ai_Connects_Us and Checkout our events: www.AiConnects.us
Dealing with schemas to give shape to IATA’s NDC initiative
Ai Editorial: Standard development is the core to the entire NDC initiative. Ritesh Gupta assesses how changes continue to take place in “end-to-end” schemas to combat issues
Is the concept of ticketing on its way out? As we move into the era of XML based concept of travelling, it looks like the industry is drifting away from ticketing and associated “multiple coupons”.
So how are changes in coding or schemas set to drive it?
This fascinating aspect lured me into finding out how executives representing different entities, including ones from airlines, are currently working on it. Working out a new technology communication standard for the distribution of air travel is riddled with challenges.
The first version of “end-to-end” schemas has already been published, but the work is being done on an ongoing basis.
Changing a schema
Issues are being discussed on an ongoing basis by a relevant IATA working group related to the management of schemas. For instance, one of the recent ones was related to a business rule document titled “Interline Flight Shopping and Order Management”.
Before we delve deeper into the details, it needs to be understood that the NDC Shopping schemas would allow airlines to distribute their full product offers and to merchandize their seats, baggage etc. And the NDC Order Management schemas are responsible for managing NDC-driven orders, from booking to fulfilment. As for the offers, they can be crafted as per the current status of a supplier’s inventory, instead of those based upon previously filed products (i.e. dynamic and personalized offers become possible). It also paves way for airlines to directly manage other aspects of the indirect distribution process.
As for the concerned group, the business and technical requirements are being met by the Distribution Data Exchange Working Group (DDXWG). It has been set up by the Passenger Distribution Group (PDG).
The main objective of this group is to document detailed business requirements and develop implementation guidance for data exchange standards in the area of airline distribution, and maintain such documentation on an on-going basis to ensure that the data exchange standards continue to reflect the industry’s business requirements.
“The group acts within the scope set by the PDG, acting as the authoritative source of the industry’s business requirements in the area of airline distribution, and under the methodological guidance of the PADIS Board responsible for the maintenance of all IATA XML and EDIFACT data exchange standards in the passenger services area,” shared a source.
“We can’t specify the number of issues (to be sorted out at this juncture), but yes the core objective is to streamline the process for all the stakeholders. From the moment a request comes in from an agency to the point the offer is tailored as per the request, it has to be ensured that the offer not only matches the intent of the buyer, but also meets the interests of the parties,” explained the source.
“We need to assess the business requirement, and ensure technical changes don’t affect any other part of the standard – technically or from standard rules perspective,” shared the source.
So coming back to the scope of the change in schema related to “Interline Flight Shopping and Order Management”, it emerged that when an order features Offer Responsible Airline (ORA) and Product Offering Airline (POA), then there should be a place holder to contain OrderID and OrderItemID of both ORA and POA. And this called for a new structure.
As to what called for a change in coding, we need to understand that theoretically, the number of suppliers in an offer and eventually an order could be numerous. But there is no need to highlight the arrangement between the parties whenever an agency is receives an offer (So if two airlines or any two suppliers come together to respond to a request, it doesn’t call for detailing the commercial dealing of the parties involved). And the current structure didn’t reflect this, it called for a change in coding.
Issue: The OrderItemID should only go to the offer responsible airline, but not back to the travel agency or the corporate agency. There is no need to show the commercial agreement or the details (mainly pricing) of each item within the OfferItemID to the buyer. So these are the sort of the issues that are being addressed.
And the work is being done by the group on an ongoing basis.
Guest Editorial by Ritesh Gupta for Triometric: How close are airlines to identifying their customers and making those unbundled offers that sell?
There are encouraging signs that this ambition is not just a pipe dream but inching closer to a real and achievable vision. Industry execs around the globe seem to be in agreement that levelling the playing field between the direct and the indirect channel and creating an infrastructure capable of handling content rich ancillary merchandising is the only way forward. Pilots are in progress and there is burgeoning interest across the globe. But these are still early days as the industry cranks up its dynamic merchandising engines and data-driven decision-making. This means letting go of a flight-centric and transaction focused mentality (and systems) in favour of a customer-centric approach with the customer journey experience at the core of what an airline retails.
What’s happening in the data world?
If we look at how the world of data analytics is shaping up, especially as the industry embraces the NDC standards based on XML, the ability to know the customer better and the context in which he is making the buying decision is quite promising. It’s like several pieces of a big puzzle getting sorted out. If the emphasis is on identifying the customer using factors of probability on the one hand or segmenting them into “look-alike” clusters on the other, then there is a great opportunity to make the most of search behaviour and match it with buying intent. This is where being able to capture and analyse the detail in the data stream between requests and offers comes in.
Analysis to determine customer insights then becomes an ongoing exercise – learning about the behaviour and relating it to the merchandising strategy, paves the way for constant refining of what to offer each time a consumer searches for his next trip.
It should be noted that the product lines of travel retailers is expanding. Competitive differences being expressed through ancillaries and bundles. Carriers are getting more creative but hampered by their existing infrastructure in the way they personalise, package, and deliver their offers via the third party channel. This is why the ability to use search insights just at the time when a potential booker is looking for something would complement the whole omni-channel shopping landscape.
“As airlines attempt to respond to a request from the agency channel, as being envisioned with the adoption of the NDC standard, it’s imperative to leverage search and booking data to deliver products and services that customers most value,” says Jonathan Boffey, SVP- Business Development at Triometric, who attended Ai’s Mega Event, held recently in San Diego.
As a specialist in the arena of travel analytics, Triometric highlights that NDC is based on XML and XML data analysis offers deep intelligence into customer shopping intent and buying behaviour.
The industry is already looking at making NDC shopping data more accessible.
For its part, Triometric has just launched an optional analytics service for airline clients that monitors the XML data stream passing through Farelogix’s NDC-Xpress platform, an offering that is equipped to deliver airline-controlled merchandising, pricing and API distribution in a SaaS model. The analytics component will enable carriers to make the most of intelligence embedded in the rich NDC-schema compliant XML message streams. The company can also deliver equivalent reporting to other NDC booking environments.
“Understanding how travellers search and book for travel is increasingly vital for airlines as they embrace NDC standards,” says Boffey. “It a big part of the NDC opportunity and it can be applied anywhere along the B2B supply chain. Whether transacted by the end consumer directly or through an agent, if the traveller data is rich enough then there is ample scope to use the available insight in the tailoring and pricing of offers,” says Boffey.
Cracking that relevant offer
“There is a big realisation (among carriers) about the potential of leveraging data sets, but as of now the use is quite limited,” says Boffey.
On the positive side, there are airlines that work with a set of rules that helps them to independently manage their own product propositions (inventory, availability, price, product biasing etc.). For example, one can work out channel-specific merchandising rules, and this would pave the way for control of what can be sold.
Bringing the element of analytics into it, by delving deeper into search requests, offers and bookings every day, one would gain an insight into customers’ travel intentions and preferences. As Triometric asserts, this would help in tailoring better offers for conversions based on customer context and market demand.
This of course depends on large data volume processing capabilities and real-time analytics, delivered via visual dashboards and key indicator alerts that give airlines visibility into shopping intent on which these more accurate and timely personalised offers depend.
So critically what would happen at the point of search?
“So on the basis of the segmentation and the defined rules that set the scene for the chosen product mix, offers are shaped up. Now an airline might have a set of 50 offers. Which ones would be most apt and match more closely to the buying intent needs to be analysed on an ongoing basis. The success of these needs to be measured – “closing the loop”. Then over a period of time the system would prioritise accordingly. This means that in responding to a search, the airline would be in a position to come up with not just a relevant offer, but one that had been tried and tested. The fact that the basket of ancillary products is increasing, makes this issue more complex. So airlines need to be spot on with their work at the backend, and gear up for the changes,” explained Boffey.
While the business information is critical, different product searches will place different demands on the booking systems and these differences need to be understood. The team at Triometric has ensured that its latest platform also gives the IT department an insight into critical operational performance, and one would be able to slice and dice data according to key metrics. These include - performance (errors, response time etc.), product availability (no offer, how many offers were given, did the offer match the search request), price sensitivity and margin (look-to-book ratios, what sort of mix is working out well – core product plus third party products etc.) and relevance/ personalization.
So what’s the timetable?
Online travel is a fast paced and highly competitive environment where systems must be monitored and maintained to run optimally at all times to avoid any degradation of service to the user or loss of revenue opportunities.
“The NDC program is gathering pace”, says Boffey, “Many carriers have been through IATA’s familiarization program and are conducting pilots. Very soon, NDC compliance will no longer be optional for all the major and most of middle tier carriers”. IATA the driving force behind NDC is already intent on taking NDC to the next level with the One Order industry-led programme aimed at modernising the multiple and rigid booking, ticketing, delivery and accounting methods with a single, flexible order management process. The ultimate goal is a single customer order record holding all data elements required for order fulfilment across the travel cycle. That’s were analytics and the intelligence it can provide is also headed.
Ai Editorial: Airlines need to address basic issues to make the most of search traffic and tapping it for ancillary revenue generation. Ai’s Ritesh Gupta studies 5 issues
Is search entirely about intent? Why it is important to assess what the user did to get to an airline’s website or a particular page? Is it time to consider Google to be an OTA?
These are some pertinent questions that airlines need to address as one tries to capitalize on traffic emanating from search. With OTAs and meta-search engines excelling in making deft use of search marketing, it is not an easy task for travel suppliers to compete.
But as a specialist in this arena, Seth Cassel, president, EveryMundo, believes airlines can forge ahead with sustainable strategies instead of trying to do SEO “transactionally”, as in “a dollar in, a dollar out.”
One of the interesting developments has been the emergence of the “Book On Google” option.
“We are not convinced that Google wants to become an OTA,” says Cassel. He adds, “Regardless, what we see right now is that organic search still drives the plurality of traffic from search engines, and that paid search is still capturing more clicks than Google Flights. That could change, but so far it has not created a fundamentally different search landscape. If anything it underscores the importance of having performance website infrastructure that can be leveraged for SEO and other direct channel marketing strategies, like paid search and email.”
We will build on this, and here are 5 issues that airlines need to address:
Cassel says not all respective passengers should necessarily have the same experience or be offered the same products as they reach an airline’s website. “Currently, there is a heavy focus on profiling individuals once they have shared information or interacted heavily with an airline, and good technology is also available to leverage the activity and information provided by the user,” he says. “Looking at the source of the traffic to the airline’s website and looking at what the user did to get to the page -- not just what the user does once they arrive at the website – can provide good insights into ancillary opportunities and that customer’s intentions with the airline.”
An airline that flies to only 10 cities, for example, cannot expect to drive millions and millions in revenue from nationwide search because it simply cannot sell that much. There are also limitations based on language, geography, and the quality of the infrastructure an airline presents to its audience. If someone is searching for things to do in Miami, for example, an airline can make an attempt to sell them a flight, but the opportunity might be very slight. Trying to get in front of a user like that is going to be expensive and limited in its ability to return revenue because there are lots of other properties willing to spend more money to get that person’s attention online.
“The flip side of that is for airlines to be aggressive about spending and growth effectiveness. Airlines should be looking at the competitive environment to see what they have to offer. Look at the website infrastructure, functionality, and online tools to drive and enhance performance. Make investments in sustainable strategies instead of trying to do SEO “transactionally”, as in “a dollar in, a dollar out.”,” says Cassel.
Ai Editorial: The onus is on airline IT companies to deliver as airlines say PSS providers are too slow in “reinventing” their PSS systems. So how to go about NDC? Ritesh Gupta finds out
The discussion around how airlines are going about embracing the new NDC standard and what’s the sort of change that is resulting in their internal system architecture is an interesting one.
Before I delve deeper into the issue of NDC implementation, it is clear that for airlines the limitations of legacy systems is turning out to be an investment decision. And it’s just not about distribution of content. For instance, airlines are being recommended to look at the entire payment chain, from gateway to risk, to acquiring, and identify areas in which they can prioritize improvements. But dealing with legacy systems isn’t easy as they have largely been designed to handle credit card transactions only.
So the going hasn’t been smooth for airline IT or e-commerce executives, especially those who are a part of FSCs, in general. There is a need to grow capabilities beyond traditional commerce management.
“The most exciting aspect about my role is to leverage changes in technology available for airline commercial activity. However, the challenge is to integrate it effectively with legacy systems such as reservation where there is no much progress in technology. Unfortunately, it is somehow still the same engine behind the scene like decades ago!” says an airline executive. A candid admission indeed.
PSS and complication interfacing
A pivotal figure when it comes to NDC is the airline PSS system. As proposed, integration of NDC specifics can be done at different levels. The possibilities are – “extend your PSS Box” or “Do it outside your PSS Box”.
An overhaul to core services (reservations, schedule and inventory, ticketing, departure control etc.) and streamlining associated functions (loyalty, revenue management, and business intelligence and revenue integrity) is a humungous task.
So how are airline commercial systems evolving?
“It seems that airline commercial systems are developing in the same direction as in the retail industry; however airline industry is few years behind retail because of the complexity of its business,” says Radoslaw Dutkowski, Director eCommerce and Ancillary Services, LOT Polish Airlines
Dutkowski believes that airlines are becoming more customer focused and are trying to effectively use the customer data which are they already have, but the processing of the data and on-request real time usage is definitely a challenge.
It is especially a problem of data integration from different systems based on completely different philosophy and aggregation methods. “I would find legacy PSS systems as a major obstacle in getting quick and well aggregated customer view as it requires complicated interfacing which may impact data accuracy and reliability,” says Dutkowski.
Fine tuning of PSS just not enough
As it is being pointed out, the fine tuning of legacy PSS wouldn’t be effective as the core of the system still lays in 70’s of the XX century.
“Legacy PSS systems should be actually redeveloped almost from scratch if it brings any major improvement, what is not going to happen in a very near future. That’s why I would rather focus on real time PSS data reliable interfacing to the external contemporary system where the proper data aggregation could be maintained. What (it) means (is) building a mirror of PSS database outside of PSS, which could be used for generation of the accurate offer for individual passengers,” explained an airline executive.
Airlines have invested considerably in several assets and want to protect these or ensure they can be easily integrated in an NDC world. Airlines are likely to have some rich content or CMS systems; however, it is likely these will have to be extended as they likely only cover the direct channel so may need to account for indirect channels also. Pricing and revenue management expertise already exist today and these systems will need to be integrated. Offer management or merchandising will likely need new investment, as the area of retailing will be a new skillset for airlines and new technologies will be required. Similarly, order management requires new investment, in order to support the lifecycle of the order across all channels used by the airline customer. Airline profile will require airlines to file their NDC capabilities so NDC actors can use this information to send NDC traffic to the airline for the end user shopping request; this will also require some investment.
A likely scenario
To mitigate the limitations of PSS systems, full service carriers or FSCs started to implement separate engines and modules such as availability and pricing, customer database, merchandising, NDC interfaces etc. outside of PSS as legacy systems are not effective enough to be used for sharpening their retailing strategy.
“Effectively it means that in the near future PSS may only be used for routine operation such us PNR creation, however the content will be fed from the different modern retailing systems,” explained a source. “ The current PSS providers are too slow in redeveloping/ reinventing their PSS system as it is capital-intensive process, so I wouldn’t expect that the major improvement in this matter in a short term. The only solution is to build the independent modularized system which will help to compete on the market, especially with LCCs as they don’t usually have such problem.”
It could be argued that a PSS vendor who is also a distributor has a conflict of interest when it comes to enabling more direct distribution, thereby putting distribution revenue (GDS segment fees) at risk. That said, NDC is a new technology and that development has to be funded, so additional fee is reasonable.
Points of contention other than technology
There are few other crucial points, as Paul Byrne, SVP Development, OpenJaw Technologies, and Member Board of Directors, Open Travel Alliance points out, that are impeding the adoption:
Ai Editorial: How astutely are you predicting what, when and where to offer a hotel room or car rental as a user is about to book a seat on a plane? Ritesh Gupta finds out
Airlines, just like other travel e-commerce sites, are vulnerable to losing out on a visitor in a matter of few seconds. It is imperative to be spot on with the sort of work that happens the moment a user lands on a site, or even when a traveller opens an email.
Airlines need to ensure they maximize any opportunity to cross-sell in order to step up the average order value. A key here is to embrace a dynamic form of merchandising and marketing, fundamentally based on data-driven decision-making.
There needs to be a mechanism that would predict what a customer is likely to buy.
“What to offer each individual customer, when, and through which channel is the new merchandising paradigm. What this means is that unique and personalized offers for individual customers based on their attributes,” says Boxever’s VP – Sales, Ultan O'Brien. There is seemingly an extreme polarization – if you get it right, then customers can buy a lot more; if your offer goes wrong, then there is cart abandonment and dismay is what a customer ends up with. “People not only buy more products when offered ones that are more relevant to them, but they are also more loyal to brands that seem to understand them and can accurately predict what products to offer them,” says O Brien.
Trip essentials – how to offer them aptly?
Making the right offer at the right time is a challenge for most, says Justin Steele, Senior Director of Innovation, Switchfly.
Predicting what would click is the key here. Here we explore how it matters:
For most airlines, as Steele says, hotel ancillaries are best sold post purchase, or after the customer has completed his or her airline ticket purchase. One great method is through a simple `pre-trip’ email. Do not just send an email with a link saying `need a hotel?’ asking the user to click a link, enter their specific dates, and conduct a search. Users can do this on their own. Instead, try to predict the top 2-3 hotels that customer is likely to purchase.
Steele further explains: Airlines can use previous customer information such as purchase history (star rating, chain affiliation), demographics, or even member tier status to begin the targeting process. Airlines can also use other factors such as length of stay, days of the week travelling, number of travellers in the party and other factors. The goal is derive the top 2-3 hotels that this customer is most likely to purchase in their destination. Then, display these 2-3 hotels directly in the email with live pricing allowing the customer to see results. Use minimal content to surround the hotel property. Large pictures with hotel name, star rating, and total price are the only 4 required pieces.
Timing is the next key. Airlines need to learn the travel journey and purchase habits of their customers. “Airlines should be able to say, “customers who book their flight 6 months in advance, and that are traveling with a family are most likely to begin their hotel search / booking process 3 months in advance” and airlines can target their email for the exact date the customer is about to think about a hotel purchase,” he says.
As for ensuring the path of booking is not riddled with unnecessary products or content, Steele says the focus should be on well-timed, planned touch points. “A certain subset of ancillaries belong in-path, varying by factors such as airline, route and business model. These ancillaries should be limited to items that a customer requires in order to complete a ticket booking,” he says. An example of this is that many customers will not book a long haul flight unless they can pick their seat. “Other ancillaries need to be sprinkled throughout non-intrusive touch points – we call these casual checkpoints. Some additional casual checkpoints include the confirmation page, confirmation emails, pre-trip emails, check in time and pre-boarding,” he says. “Understanding the relevance of the offer and the timing of the offer in the customers purchased decision process can help airlines determine which products to market, when to market and where to market. Then tweak and test. This is definitely not a ‘set it and forget it process.’ The process needs to continue to be refined by the airline.”
O'Brien says the existing descriptive and diagnostic analytics landscape has traditionally been a very lean-back approach to merchandising ancillaries. A typical example might be to create an email segment from an operational CRM by querying all family bookings who purchased previous (the same time last year), and who added a specific ancillary (also bought a hotel, car or insurance package). “It’s using historical data points to inform a contact strategy with the customer that is delivering increasingly limited success – mainly because it is generally based on stale information and is part of a generic grouped campaign rather than personalized to an individual,” he says.
He adds, “The result is a form of scatter gun marketing communication – with a kind of “throw mud at the wall and see what sticks approach” – which is batched rather than real-time (so tends to be stale and out of date) retrospective in its nature and based on historical information (thus missing any intermediate interactions the customer has with the brand that can indicate an intent to spend, a buying persona, a context and the relevant product(s) of interest).
The New World
By using a data-first merchandising platform - which captures all interactions as well as shopping and buying behaviour across all channels – airlines can commence building a unique contact strategy for each visitor, traveller or customer.
“The use of predictive analytics to drive offers in practical terms means you stop asking the question “What products do I need to sell” and start asking “When will a specific customer next make a purchase?" and "How best can we communicate to them to make it happen?”, says O Brien.
According to him, predictive marketing is really about understanding who the customer is, what products they might be interested in based on their behaviour as well as their transaction history – and then marketing contextual relevant personalized offers in an automated way, orchestrating this across all customer touch points to achieve what is referred to as customer-centric marketing. So, when a customer has not purchased a product it’s then about inspiring them or pre-targeting them with appropriate travel experience. When they are however in the purchasing funnel it’s about offering the optimal assembled product set (or dynamic package) to convert them, and re-targeting them with the next best ancillary to widen their traditional travel basket.
Whatever airlines strive to offer needs to be spot on. Anticipating the next product a customer will want to purchase right before the customer does, then making it available to the customer through the proper channel is a top priority right now.
Passengers see different products, different prices, different offers or bundles. Airlines are trying to overcome such inconsistency.
We all avoid irrelevant offers. It could be that the offer isn’t fit for that moment or just isn’t for me at all.
If I receive an offer that is actually invalid, not just uninteresting, it can be quite annoying.
So it could be a product that I could buy but just don’t want. As for the “invalid” part, it could mean there really is a strong reason, based on the data the airline has, that I will never buy that product. For example, I won't buy it because I have already bought it. Another example is that I’m already entitled to that upgrade or lounge access for free because of my tier.
“It is imperative for any airline to having intelligent filters and business rules in place - to do things like remove invalid offers are an essential first step before you can implement recommendations based on data analysis,” says Mark Lenahan, VP of Product Strategy at OpenJaw.
E-commerce entities, including retailers, today assert that customers do not see channels. Whatever is offered, wherever it’s seen - needs to be useful or inspirational for a future buy.
But airlines have some specific challenges.
If we were to talk whether airlines are in control of what they would like to offer or not, Lenahan says there currently isn’t enough control because of both technical and commercial aspects.
“Airlines, in general, feel that their current technology does not provide sufficient control over what they sell. Most airline environments consist of a mix of technology vendors, but some of that technology is very old and inflexible,” he says.
There are also commercial issues with the way airline product is distributed, for example, there is no such thing as a gentle transition from being “full content” (meaning all fares on the GDS) to having channel-specific pricing. The airlines either live with the status quo or start what gets described as a “revolt”, as we’ve seen with some carriers in recent months.
The combined impact of complexity in both technology and distribution leads to inconsistency for the passenger – and their overall experience.
“The passenger sees different products, different prices, different offers or bundles, all depending on which site they are visiting, or what device they are using,” says Lenahan.
“If an airline goes to the effort of sourcing or creating a unique product, whether it’s an on-board experience, destination event or hotel package, the decision on where and how to sell that product should be based on what the passenger wants and commercial needs of the airline, not on what channel (or silo) the passenger happens to be interacting with at that precise moment,” says Lenahan.
Getting better at retailing
So technology and infrastructure is definitely one core aspect of ensuring a flyer is offered something that is likely to click with him or her. But this wouldn’t be possible in case airlines continue to stick to non-integrated pricing and merchandising solution providers (these entities can have competing priorities, too). Airlines need to avoid going for multiple merchandising processes/ systems. Managing it across direct and indirect channels is neither scalable, nor technically, economically, or commercially feasible.
The second aspect is data analytics for keeping a tab on the intent of the customer. “I think it’s still early days,” says Lenahan, referring to the maturity of the same.
Delving deeper, he says “inspiration” (meaning capturing the customer’s attention, getting them to shop, or inspiring them to look at a destination) is a source of data (e.g., click through, impressions) but marketing need data to begin with. Search and other datasets from external sources, i.e., Google, meta-search etc., can help airline marketing teams to better understand what to promote, where to target their content curation and allocate digital marketing budgets.
Referring to the concept of retailing experience, Lenahan says, “(At OpenJaw), we try to help inspirational shopping by aggregating live product data such as compelling content, accurate availability and live pricing from multiple sources, for all product types, into one platform.”
This supports inspiration and “browsing”, selling the outcome, focus on the destination, and better customer acquisition.
Permutations and combinations based on data
Continuing further, Lenahan mentioned that after a customer has narrowed down to a specific destination, data helps and informs about priorities for product biasing, bundling, etc.
In theory if you have 10 ancillary products, you have 45 possible 2-product bundles, or 120 possible 3-product bundles. However, many of these bundles don’t make logical sense and in any case, you still have to let the customer choose any of the 10 products individually. While that’s only partly a data problem, it’s also a user experience and design issue. “There’s no doubt that data analytics and behavioural economics can help. The challenge of recommending the 1st, 2nd, 3rd and so on from 200 hotels is quite different,” explained Lenahan.
For their part, OTAs believe it’s only after a considerable number of years they have reached a stage where they can rely on repository of data to make ensure whatever is offered sustains the booking flow. For instance, as a B2B specialist, RentalCars Connect, relies on the intelligence that its B2C business has its disposal, and ensures the products offered for a certain destination are likely to be bought.
Equally important is being adept at statistical demand forecasting models to predict demand for ancillary products. Also, price optimization is another component that needs to be factored in.
Also, as it turns out, some automated processes aren’t equipped to come up with precise recommendations when it comes to personalisation. Machine learning algorithms are playing their part, by evaluating which offers were most valuable. If an offer didn’t click with a passenger, the algorithm can capitalize on that intelligence and integrate it to work out for relatively more accurate offers.
Overall, travel retailing, with a broad set of products, requires a broad set of tools. Also human curated product management will not go away, even if machine learning is used to automate some aspects of bundling or pricing, asserts Lenahan.
By Ritesh Gupta
Gain an insight into the world of retailing at Mega Event in San Diego which is taking place on the 4/5th of November. More information at www.MegaEvent15.com
Follow Ai on Twitter: @Ai_Connects_Us and Checkout our events at: www.AiConnects.us
Tracking users who have deserted a travel ecommerce shopping cart and approaching them with a relevant message or ad is becoming common. But doing the same with a seemingly blind targeting tactic is dangerous.
There is something wrong with the current level of personalized retargeting.
A recent experience exemplifies this. I accessed a car rental OTA website and checked options for moving around in London. The site stood out for its functionality, especially the mapping feature and the ease with which one could compare options. All in all, it was a pleasant experience, but I chose to leave the site before booking.
After doing so I was bombarded with a display ad, session after session, encouraging me to click and book the rental car that I searched for. There was no modification in the content of the ad at any stage; the ad formats just changed depending upon the site that was being accessed. Quite annoying to say the least.
So if travel advertisers are attempting to capitalize on the identity of the user and their search behavior, but are only re-approaching users with the same cart content, conversion value or even the opportunity to upsell can go awry.
“It is a clear case of spam,” says Dave O’Flanagan, co-founder and CEO of Boxever.
“If done intelligently, re-marketing can serve the intended purpose of directing a traveller back to the website which can drive conversion, but blanket retargeting without any intelligence applied definitely presents the risk of annoying a customer and potentially driving them to a competitor,” he says.
Personalisation isn’t only about conversion and revenue optimization, but also about experience optimization. A personalized experience can create positive sentiment and increase brand loyalty.
Identifying the right person and the customer that is most likely to buy are just two components of personalisation. “How the ad served is affecting a particular brand also needs to be assessed, as well as monitoring responses to ads in order to determine its effectiveness (too needs to be done),” said O’Flanagan, who acknowledged that the travel industry is in the early stages of serving content that aligns with the interests and activity of the user.
“It’s also important to know when not to retarget, as unnecessary targeting can annoy a customer. Knowing when not to do it requires a certain level of sophistication,” he added. There have been instances where customers end up complaining that their booking flow is interrupted or disturbed by unsolicited ads and messages on a travel supplier website.
“What's important in these scenarios is understanding which customers will be most responsive to that tactic - and at what saturation - based on propensity scoring, past behavior, current context, and so on. In some cases the best action may be to take no action at all, and save the re-marketing dollars for a better prospect and ensure that you don’t annoy the customer,” explained O’Flanagan.
Count on personalisation as a journey
It’s important to view personalisation as a journey, not a single moment in time.
Airlines are in the midst understanding the whole game.
In a recent interview, Nik Laming, general manager, loyalty division at Philippines’ leading carrier Cebu Pacific Air, shared that identifying customers across proliferating social platforms, devices and channels is difficult and becoming more difficult every year.
So how should airlines go about building up resources and infrastructure in a practical way to address these concerns?
O’Flanagan, who agrees with Laming, says airlines, OTAs, and hotels have been awash in traveller data for years; the challenge is pulling it all together into a meaningful profile that can be acted upon with intelligence and recommendations. “It's important to define what you want the end goal to look like (e.g., 1-to-1 offers across all digital and offline channels), and then build toward that in phases. For example, perhaps you start by merging web site data with email data, and then layer on customer service history and social data, to create a more robust customer profile. Each time more information is added, it creates new opportunities to understand your customer, refine segmentation, messaging, and offers across one or more marketing channels,” summed up O’Flanagan.
Travel companies are certainly beginning to infuse more personalization into the booking process and beyond.
Referring to one moment of surprise and delight, O’Flanagan shared a recent experience: “I travelled with Air New Zealand (a Boxever client) recently and when I arrived in the airport their app asked what type of coffee I would like in the lounge. So I selected a cappuccino and made my way through security. Once I made it to the lounge, the barista had my cappuccino waiting on the counter with my name written on it. It was a complete surprise, seamlessly blending the online and offline worlds into a delightful customer experience.”
Sounds exciting indeed.
By Ritesh Gupta
Learn more about personalisation at Mega Event in San Diego which is taking place on the 4/5th of November. More information at www.MegaEvent15.com
Follow Ai on Twitter: @Ai_Connects_Us and Checkout our events at: www.AiConnects.us
Optimizing merchandising on an airline’s website is not rocket science but it is easier to get it wrong than it is to get it right, says Rentalcars Connect’s Ady Guthrie
In how many clicks a travel booking should be completed? How much time would it take to do so?
These might sound like simple questions, but airlines continue to learn and test various designs and functionalities as they try to optimize the booking flow in order to optimize the ancillary revenue generation. Airline e-commerce executives dread the idea of poor execution of merchandising, be it for air ancillaries or non-air ancillaries. For them, of utmost importance is to offer an enticing design, aid users in the task they are about to perform, and eventually present them with results that are relevant and fine enough to complete a booking.
As an experienced professional, Ady Guthrie, Global Director – Strategic Partnerships, Rentalcars Connect says a key would be to get the basics right first and then build more sophisticated cross sell options down the line.
“It’s all about using the data you have to offer a relevant car at a relevant destination. It’s not rocket science but it is easier to get it wrong than it is to get it right, which some do. We’ve have spent years building the relevant B2C data and analyzing this to understand how to properly execute this throughout our partner integrations,” he says.
“Our platform is predominantly search-driven, with destination information directing the results. So where the customer is searching from and their eventual destination shapes up the search results. Sometimes previous searches aren’t that relevant, and if irrelevant results are shaped up by something that isn’t actually the intent of the customer, then it can hamper the user experience and the performance of the site. Our objective is to facilitate simplest and easiest user experience, and pave the way for conversion,” shared Guthrie.
So the focus is on offering relevant products and facilitating a quick purchase, be it for Rentalcars B2C (rentalcars.com) or B2B platforms (rentalcars connect).
There is talk of real-time personalisation on the web. Merchants today are in the process of setting up an interaction timeline for each visitor and customer from the instance they visit a website. Once they have an “identifying” event then all of the rich anonymous behavior is incorporated into the “known” profile. And then known and unknown site visitors can be served dynamic content. Guthrie acknowledged such developments, and says the level of sophistication also depends upon the industry. “I might log on to Amazon.com 50 times in a year, but a car rental site only few times in a year,” he mentioned, referring to apt utilization of resources. “But yes, destination content needs to be relevant, and there are certain areas such as these that we are looking at as well.”
Sustaining the booking flow
Patrick Dowling, Regional Director, Rentalcars Connect says the team tries to ensure its merchandising offers a seamless booking flow, along with a certain level of personalisation.
Citing an example, Dowling mentioned that if there is a PNR featuring a certain destination and it emerges that there are four people with four bags in total, it doesn’t make any sense to offer a car that won’t able to serve the basic needs of the journey. Other than relevancy, the design and functionality that at times are seemingly “cool” can actually derail the booking process. Guthrie cautions against “over-engineering” the whole exercise, and strongly recommends a quick response rate to testing in order to embrace the best features and design on the site.
It is also important to garner the attention of a flyer at the right time.
“(This can be done) partially by testing but also by building an understanding of which ancillaries convert for which destinations and for which customers. Also, if you’ve offered a car twice already for example, then logic would dictate to offer another ancillary product to maximise the opportunity of earning some revenue even if it might not always be the most lucrative,” shared Dowling.
Fine-tuning the product
Sometimes even the intention of a supplier isn’t completely understood or customers just shy away from sharing extra information.
Citing an example, Guthrie mentioned that when customers were being asked to share their flight information in order to ensure that their car is available or waiting even if there is a delay, users were dropping out from the booking flow. The team at Rentalcars reassessed the situation and chose to introduce additional functionality to elicit such information. “We asked customers departure and arrival dates, and then there was a drop down to dynamically choose the flight number from the list of flights,” shared Guthrie.
Handling partner integrations
The company is certainly counting on its expertise it has developed from its B2C brand, RentalCars.com.
“We work with airlines partners to ensure the products offered for a certain destination are likely to be bought. For low conversion rate destinations, the cross –sell products could be insurance, hotels, extra luggage etc,” mentioned Guthrie. Rentalcars Connect banks upon the repository of data and the intelligence that its B2C business has its disposal, and this also paves way for prudent product development as part of ancillary offerings. Guthrie also mentioned that the eventually this plays a strong part in any co-brand solution as well.
A case in point what needs to be shown to a Chinese customer in order to book a car rental is going to be different from a U. S.-based user. As the team at Rentalcars experienced, one of the risks is that Chinese customers may not fully understand what they are and aren’t covered for when they drive in another country such as the US. So in case there is an accident with another vehicle and the occupants are injured, those legal claims can easily run into a hefty amount, unlike what Chinese customers would generally expect. So rentalcars.com has developed this know-how, and this in turn improves the conversion rate when the B2B division takes over the booking flow from its airline or any travel supplier.
“As we take all the learnings from rentalcars.com, where we conduct endless and robust testing and experimentation on both the booking flow and pricing (front end customer experience), we only roll out positive margin and conversion improvements to our partner solutions,” mentioned Guthrie.
He also referred to another crucial aspect.
“We try not to interfere with any partner integrations that have been tried and tested on rentalcars.com – why would we if it may negatively affect conversion and margin! We don’t change the platform that has been tried and tested but we do collaborate and tailor the integration based on the partners preference or requirements, we do believe in collaboration,” mentioned Guthrie.
So what’s the benchmark for success that an airline partner should typically look for – average order value or conversion rate when they look at the efficacy of a car rental partner/ solution in overall bookings?
Guthrie says the same varies depending on an airlines source markets and destination mix “but we nominally see PNR attach rates of between 1-4% and the conversion rates on our partner solutions tend to be higher than that on rentalcars.com as the customer is in a more purchase ready frame of mind”.
By Ritesh Gupta
Latest developments pertaining to personalization will be discussed at our global Mega Event (04 & 05 November 2015, San Diego).
Follow Ai on Twitter: @Ai_Connects_Us and Checkout our events at: www.AiConnects.us
IT infrastructure and distribution of data - where does NDC fit in today?
NDC, as a communication standard, has been in the limelight for three years or so. Ai’s Ritesh Gupta finds out how this initiative is shaping up on a couple of counts
There has been a continuous scrutiny of how the XML-based data transmission standard is going to pave way for superior air travel retailing.
Standard development is the core to the entire initiative. The first version of “end-to-end” schemas was published late last year. Airline adoption, from testing to deployment, is expected to be gradual this year as well as in 2016.
As much as airlines themselves refer to the limitations of legacy distribution and the potential of a modern industry communication standard in opening gates for a differentiated product offering, they, too, have comprehended a couple of lessons over the past few months. For instance, one can’t really foresee replacement of all the airline systems with NDC-compliant systems at this juncture. (It [replacement] is still untimely, as there isn’t yet a standard to define “NDC-compliant”,” mentioned an airline executive. Also, there is a need to gear up for the data that NDC will bring along with it. Indeed there is an option to enhance offers and corresponding take rates. This is going to be a key development for airlines in how they market their product.
Here we explore a couple of important aspects that are related to IT infrastructure and the impact on distribution of data that NDC is resulting in:
To their credit, carriers have invested in merchandising platforms tailored to dynamically bundle ancillary services based on a passenger’s value. There have been concerns about transparency, with questions over whether the price resulting from a specific request is lowest or not, in case pre-filed fares come to an end. Also, multiple direct connects featuring agencies can make price comparison more difficult.
For their part, GDS companies have been integrating with NDC-like XML APIs from carriers.
At the same time, a section of the industry also points out that NDC-XML is only a part of the merchandising effort underway in the industry and is far from the only connectivity standard.
“We are already seeing that usage of NDC -XML by airlines and GDSs will vary in its shape and form, resulting in a mix of EDIFACT and XML connectivity,” shared a source.
A case in point, IT specialists are introducing their merchandising systems, integrated with their PSS offering. They promise that this would also help carriers to work out dynamic and relevant offerings: any time, any point of sale or channel, direct or indirect, through any device.
So is this just another way of improvising existing legacy technology? Can it last? Only time will tell.
It is often debated whether the IT specialists behind PSS’ deliberately tend to resist change and rather not succumb to an attempt by new players to enter their territory in any fashion.
Sebastien Touraine, Head - Airline Merchandising, IATA, says the NDC technical standard provides the opportunity to implement a modular environment where valuable assets are leveraged, out-of-date assets are renovated and new assets can be added easily.
“In effect, this modular architecture will allow airlines to respond rapidly to changes in their business environment; this is a must for retail organizations,” he says.
But it’s time airlines move on to an API or other XML-based technology. It’s simply for the fact that even “with the frequent updates to and from the ATPCO databases, the pricing remains relatively static and is difficult to personalize”, as an executive shared. The IATA NDC initiative is expected to improve this.
It must be mentioned that airlines have been waiting for the continued rollout of EMD-A, which makes it possible to relate a non-flight item to a flight ticket, offering insight into the buying trends of flyers. Until EMD-A comes to full implementation, any ancillary sales will continue to be recognized as a separate transaction; however, that does not prevent an airline from selling it, it only makes it difficult to associate it to a specific flight activity.
The industry has been waiting for standardization of ancillary offerings to support interline sales, fulfillment, settlement and reporting, etc.
Touraine says: “Having led on behalf of IATA, the Electronic Miscellaneous Document (EMD) implementation across the industry, modernization of end-to-end process, capability and industry standard is a must. While some industry functions will be simplified/ modernized through the capabilities enabled by the NDC standard, there are definitely more opportunities to update/ modernize the order and delivery processes. As airline merchandizing provides greater options, some processes (for example PNRs, ETs) were not designed with this level of complexity in mind.”
He also referred to One Order, an industry-led initiative intended to “modernize the multiple and rigid booking, ticketing, delivery and accounting methods with a single, flexible order management process”. As a result, a flyer wouldn’t require different reference numbers and documents. One order reference would be sufficient. For carriers, this would do away with expensive reconciliation between PNRs, e-tickets and EMDs. And agencies would be able to pursue same procedure to book flights and products from airlines regardless of the airline’s business model or technology capability.
According to the IATA, the One Order business case is ready and will be presented to the Board of Governors in December this year for the final decision on whether to move forward.
Going forward it’s interesting to assess how airlines can get closer to creating their offers. Also, the NDC adoption is the way to go, but how much of IT infrastructure would be changed hangs in balance at this stage.
Gain an insight into the world of NDC at Mega Event in San Diego which is taking place on the 4/5th of November. More information at www.MegaEvent15.com
Follow Ai on Twitter: @Ai_Connects_Us and Checkout our events at: www.AiConnects.us