The personalization of travel: Connecting human and digital experience through data and machine learning by Jasmin Schmidt-Stiebitz
Jasmin Schmidt-Stiebitz
Head of Data Product Steering (Personalized interactions)Personalization of Travel by Lufthansa Group
Welcome to an informative session led by Jasmine Schmitz, the head of data product steering at Lufthansa Group. She illuminates how Lufthansa, a leading German airline, personalizes travel experiences for their customers by leveraging data and implementing cutting-edge techniques.
About Jasmine Schmitz and Lufthansa Group Digital Hunger
Jasmine has been a part of the Lufthansa Group for five years now, contributing to various departments. Her current role involves steering the data product for personalization to create an optimal journey for customers across digital, human, and physical touchpoints. The Digital Hunger entity within the Lufthansa Group, which Jasmine helped establish, unites all departments and teams that focus on digital interactions for customers across affiliated airlines, like Austrian Airlines, Lufthansa Airlines, Swiss Airlines, and Brussels Airlines.
Data Enhanced Customer Interactions
In 2022, Lufthansa served over 100 million passengers, emphasizing the vast potential of personalized interactions. The ultimate goal is to create satisfied and loyal customers by providing a connected experience that combines physical features, human interactions, and digital solutions. To achieve this, Lufthansa largely relies on data-enhanced customer interactions, which Jasmine explains in detail.
Personalization on Digital Touchpoints
Netflix's personalized movie recommendations and Amazon's product suggestions are excellent examples of digital touchpoints. In a similar vein, Lufthansa employs machine learning mechanisms on digital touchpoints to stimulate customer inspiration for new destinations and offer relevant products. Jasmine provided two examples of how machine learning is currently applied.
Destination Inspiration
In the first example, a machine learning model uses customer location and timing data to suggest potential travel destinations. This model currently treats people located in the same city similarly. To make it more customer-specific, it will soon consider individual characteristics and personal interests to make the recommendations more accurate and targeted.
Pre-flight Email Recommendations
The second example revolves around an ancillary recommendation sent via email during the pre-flight stage. This model predicts the interest of each customer in different items and sorts them based on that prediction, pushing the most relevant content to the customer. The same model is applied to both emails and webpage touchpoints to ensure a streamlined experience. Current efforts are being carried out to fine-tune this model.
Human Interactions Onboard
Jasmine also elucidated how data about customers is brought to crew members to personalize interactions, especially on board, despite the challenging digital environment. One intriguing feature is a scoring system that predicts customer satisfaction levels, allowing cabin crew members to address any discontent proactively and improve the experience.
The Role of Customer Profiles in Personalization
A crucial element of personalizing a customer's journey is having access to a detailed customer profile. With customer permission, Lufthansa captures and analyzes data over time to improve its scoring methods. Besides predicting dissatisfaction, another scoring example involves calculating a customer's potential lifetime value, i.e., future spending behavior. This data helps enhance internal steering, making it possible to offer special treatment to high-value customers.
The Future of Personalized Experiences at Lufthansa
In the continuous quest to enrich customer experiences, Lufthansa is exploring ways to incorporate customer preferences and requirements into their profile, such as dietary preferences and language of communication. While physical preferences may not be as crucial for digital interfaces, it becomes invaluable for human touchpoints like cabin crew or service centers.
Active ideation efforts, including staff interviews, customer surveys, and workshops, thrive to identify more relevant data points to personalize the human experience further.
Conclusion
From utilizing machine learning to predicting passenger needs, Lufthansa is pioneering the personalized travel experience. There's an ongoing focus on optimizing methods and models to ensure customers enjoy a seamless, connected journey. Measuring these impacts is also a priority. Even though quantifying this impact becomes complex due to various influencing factors, attribution logic helps realize an uplift in sales due to tailored product offerings.
Though challenges exist, the future of personalized experiences at Lufthansa Group promisingly hints at improvements in app usage and providing for specific customer groups such as elderly and differently-abled passengers. If you have any ideas or thoughts, feel free to share and contribute to the continuous enhancement of personalized travel experiences.
Video Transcription
Hello, everyone. Welcome to this session. I think people are already joining in the in the waiting room. Hi. Uh We have the pleasure of having here today. Jasmine Schmitz. I hope that I pronounced the name correctly.She is the head of data product steering at Lufthansa digital uh digital handler um in Germany. And she will be talking more up to today about the personalization of travel. She has a very exciting session coming up, talking more about uh applying customers course and human interactions.
And generally we're really looking forward to hearing more from you whenever you're ready, feel free to take the lead.
Thank you for the introduction. I'm happy to speak today about the personalization of travel and give you a few insights how we at Lufthansa Group and especially at the Lufthansa Group Digital Hunger are applying the data that we have about our customers to create an experience and customer experience that is personalized and uh the perfect experience for our customers.
A few words about myself. Uh My name is Jasmine. I am the head of data product steering and that means that I am within the Lufthansa Group Digital Hunger Responsible for personalization. So for creating the best possible journey for customers on digital human and physical touch points.
I have been with Lufthansa Group for five years now in different departments. And since last year, I was uh involved in building up the new entity within Lufthansa Group. The Digital Hunger. There's a few words about what the digital uh the digital hunger actually is. So we are part of Lufthansa group which includes multiple airlines. Um for example, Austrian Airlines, Lufthansa Airlines, Swiss Airlines, and Brussels Airlines. And uh with the digital hunger that we created last year, we have united all um departments and teams that working on digital interactions for customers with those airlines. That includes, for example, airline apps, airline, web pages, the booking flow and all different uh ways how customers can interact in a digital way with the Lufthansa Group Airlines. It doesn't only um include the digital interactions, but it also includes data enhanced customer interactions.
And today in this lot, I want to tell you more about what does this actually mean to create data enhanced customer interactions? A bit of background. In 2022 we had over 100 million passengers that flew with our airlines. So our reach with personalized interactions can be quite big if we managed to um to gather the data and reach out to those customers for actually personalizing their experience. Our goal is to create happy customers so that in the future they will come back to our airlines and be loyal customers and um fly with Lufthansa Group in the future as well. Why are we actually um putting such a strong focus on personalization of multiple touch points? We believe that uh as a premium airline, we need to offer our customers a connected experience that really unites the physical features that we have as airline group, the human experience that we can offer by our crews on board, by our airline staff at the gates at the lounges, for example, and digital solutions only when combining those three different uh ways of communicating with our customers, our customers really feel like they have a connected journey, they are known and we can appreciate our customers with their individual background and in their individual situation, we wanna make sure that we make things easy for our customers and empower them to give them the right tools to um to create their own journey, their own experience and to also empower our employees to um create a personal experience for our customers by knowing things about them and by giving them the tools to interact individually with every person that they see on board.
So how do we actually do this on digital touch points you may be familiar with, for example, Netflix, personalizing um what what kind of movies they offer you or Amazon uh showing you the products that you might be interested in and or digital touch points we use quite similar approaches to get our customers inspired about new destinations, for example, or to um push the right ancillary products or things that they would like to buy on top of their already existing flight booking towards them.
I brought two examples. Are we at the moment, apply machine learning mechanisms on digital touch points to reach those goals of inspiring our customers and um providing them the relevant uh products for them. On the left side, you see one interaction that is that we currently have, which is our destination inspiration interact here, we have a machine learning model that's currently using the data that we have about customers location and the timing that they come to our web pages.
And based on that we saw them uh locations or destinations they might be interested in flying in based on what other customers from this um from this origin. But at this time, we're interested in and uh what we know about this customer at the moment, this model is um very or treats people at the same time in the same um place in the same city quite similarly. So at the moment, we are working on improving this model so that it really takes more characteristics about specific customers with that after they have logged in, for example, about their interests, for example, um one person might always around Easter time, book a destination for a beach trip.
And we, we are currently working on taking this information and implementing it into the model so that the um recommendations about destinations get even more specific for this customer. This is one example of how we have a model life at the moment. And we are um using additional data and implementing additional data to improve this model and inspire more customers about flying to new destinations on the right side. Um I brought an example about our ancillary recommendation that we have at the moment. This is for example, an email that customers get after they have already booked. So they are what we call in the pre flight phase. They have an open booking, their trip is coming up and they might be interested in specific information or in specific products. For example, a customer is traveling with the child and they may wonder. Ok. Um Do I need to uh take special measures so that my child is um accompanied the right way on the trip or um a customer is entitled to go to a lounge? What lounge do they actually want to go to? And what kind of information do they need in this machine learning model? We implement both information items like what you need to know for traveling with Children and um product information like now preorder your food uh for the next trip or maybe you're interested in an upgrade and create specific scoring for every of those items and this specific customer, how much would they be interested in the scoring based on the scoring of different products, we then sort the different info items and product items in um the best sorting for the specific customers.
And in this email, for example, push the three most relevant contents to the customer. We have the same model at the moment uh live both in emails and on the web pages. So on the different touch points, customers can see either three items in the mail or six items on the web page. And we are currently working on approving the model so that it's not. Did the customer, for example, yesterday, look at the web page and was interested in one specific product. And now today they're receiving a mail and they may be interested in this product again or they didn't like it. We are going to push a different product because we think they they would be more interested in that this is another step of model improvement that we're working at the moment to make those digital touch points even more personalized. But as I mentioned before, uh personalizing digit, digital touch points is not um what sets us as an airline group really apart from other airlines or from other digital players. And we want to combine the digital interactions and the data that we have also with human interaction, human interactions as an airline happen and various touch points.
But one of the main touch points where customers interact with humans is on board with cabin crew, for example, on board, we have quite a specific environment. Usually there's no internet connection or at least no super stable internet connection that can consume big loads of data.
So um we are currently working on bringing data about specific customers to crew members who are in direct contact with our customers. I wrote one example of a scoring that we have um we are implementing at the moment where we score for every pass that gives us their permission. How dissatisfied are they at the moment? For example, based on cancellations that they had in the past, based on irregularities like lost baggage, like delays, things like that. And for each customer, we then calculate. Ok. Is this customer probably happy with that experience or are they super dissatisfied? And we should really take action to um reach out to them and to make an effort so that they become more satisfied. Again, this information is then pushed to the crew device and the crew can then see in the seat map that you see on the right side in this cabin. What kind of customers do I have? And oh, there's one who's probably suited super dissatisfied, they can then reach out to the customer and apologize for the experience.
Uh say thank you that they came back again and maybe also offer a gift or offer a better seating option depending on availability to create a special moment and a special experience. For the customer today. This is just one example of how um scoring could also um improve the personal experience that customers have. And we are at the moment working on creating more and more of those different scoring of those different preferences to en enhance the journey of our customers on board.
The entry way for this is always the customer profile because we need the permission of our customers to be able to store their data, to be able to analyze the data and also to be able to reach out to them. So their profile is really the base for us on how we can uh personalize the journey of our customers. Once we have the customer profile and we have some data from the customer profile, we can then create scoring like the dissatisfaction scoring when we really look at what happens to this customer over time. For example, in this case, over six months or also how does this customer behave in the past? And what kind of um behavior do we expect for the from the customer in the future? Another example is the customer lifetime value where we calculate, what kind of behavior do we expect from this customer in the future? What kind of spending do we expect from the customer in the future? And is this a super valuable customer for us or is this a customer that's maybe that or that we don't expect to come back too often in the future.
This customer lifetime value is another example on uh how we could use this data also for internal steering because for example, when we have an aircraft change and uh people need to be upgraded to business class. For example, we should look at which specific customers have a high value and which specific customers maybe we can really make uh make an impact on when treating them in a special way. So those uh scoring we are at the moment implementing at uh human touch points and also at internal steering uh mechanisms to create a better experience for our customers. Those currents are only one part of the picture though it could be in the future, also enhanced with customer preferences. So like does this customer um what's the native language of this customer? Does this customer always eat vegan meals? Does this customer always drink red wine when they're on long haul flights? All this different information we want to um bring to connect to the customer profile and then bring to the touch points where it actually matters. It doesn't or not all of those scoring matter at all touch points. For example, in the cabin crew, it doesn't really matter. Um or for the cabin crew, it may not really matter what the future spending behavior of a customer is.
They may be more interested in what they like to do on board or what are their preferences on food or on drinks? For example, or what language to approach them with. But it's a different picture in the service centers where uh the customers call to have some complaints or to have some open issues that would be very valuable to also know what is their future value and how do we treat the specific customer at the moment. We are regarding this in a continuous ideation stage. So we are conducting staff interviews. We're conducting customer service, we're conducting ideation workshops to really find out what kind of information do the colleagues need who are talking with customers every day, who are in direct contact to be able to personalize the human experience as well. And any additional uh ideas on this are highly appreciated so that we can use um or we can personalize the human and the digital experience in the future. So those are some examples of what we do at Digital Hunger. And um like I just mentioned the collaboration mode at the moment also consists of um doing design interviews, design prints and finding out what are the things that we really want to deliver in the future so that we bring the best possible experience to our customers on digital human and also on physical touch points.
Oh, I see. Um one question in the chat already. I think we have a few more minutes time for questions and uh maybe discussion items that you would be uh interested in. And um yeah, I'm looking forward to hear about your questions and your ID is I see one question already. Uh is Lufthansa using machine learning for pricing tickets or is it purely for customer experience for us? The focus at the moment is on customer experience in the other uh departments in pricing and network planning. Of course, they are also working on using um additional data sources to learn more about roots, to learn more about customers. But this machine learning focuses at the moment uh really on the customer experience and creating a personalized experience of different touch points. Then next question, how can machine learning algorithms be used to predict and anticipate the needs and preferences of individual travelers?
Yeah, that's actually a really good question. And this is what we ask ourselves in our daily work as well and what we are working on. So on the one hand, the needs and preferences, um the one example that I talk about the dissatisfaction, this is one algorithm that we have to predict. What may this customer need? Special attention, special service and apology, some loyal organization measures, but also um their product affinity.
So what kind of information item do they need? That's what we um build the machine learning algorithms for to anticipate this. We use um information on what other customers needed as a reference here. So uh for example, people traveling with Children always click the uh what kind of food can I preorder um option? And uh we would then push this espe uh especially to people who are traveling with Children. But this is really what we are um building our machine learning algorithms for to predict what information to our customers need and what service options do our
customers need. There's also one question in the Q and A section is is there any quantifiable impact so far by implementing these changes?
Yes. Uh We are also measuring the impact that we're making. For example, with doing specific um product recommendations, it's not always easy to quantify because we need some kind of attribution logic to find out which purchase was now done because of this machine learning algorithms.
For example, or because this customer saw this specific um email and this specific communication measure. But we have uh an attribution logic in place. And uh we see that in the format where we have the um machine learning model predict the um the affinity for specific products that there is a bigger chance of the customer buying the product than if it's just a static um a static implementation of products. There's many things that are actually influencing this one. The biggest one is one product that we recommend the upgrade costs probably over €100. While a different product that we recommend the food preorder option only costs €10. So it's sometimes a bit hard to measure really what is uh what is the end value or to compare the ends uh revenue value by this. But this is what we are also working on and um measuring the machine learning model uplift is very essential for us. Yeah,
perfect. I think there's two more questions. The first one is is there any specific prediction for different abled and elderly travelers?
Mm We don't have specific recommendations for specific groups of people. So all of our um algorithms and scoring mechanisms at the moment, look at all kinds of travelers who gave us their permission no matter what their ability status or their um their age, for example is. And so far, we haven't found a use case where that would make a huge difference. If you have an idea, uh we would be very happy to look into this. So I have an idea. If you have an idea, please go ahead and share because we're always interested in uh also making various customer groups be more represented and prioritized in the experience.
Perfect. Um And another question is, did we improve the Lufthansa app usage compared to marketplaces after the usage of machine learning?
Can you repeat this question? I'm not sure if it,
did we improve the Lufthansa app usage compared to marketplaces after the the usage of machine learning?
Mm We can't really measure that at the moment because in the Lufthansa Group app, I'm not sure if you're familiar with the latest developments. We just launched a new app for Lufthansa Airline and also for the other airlines within Lufthansa Group. And um so far, we haven't really implemented the machine learning models in the app because it was just newly launched in the future though, uh we also always look at the reach effects on how many customers are using our apps. And of course, we hope that by personalizing it more, we will also get more customers to be to come to our platforms and to interact directly with us as Luhan. This style of interaction is in, in general, very important for us. And one of our main goals also when personalizing our digital charge points.
Nice. I think uh that's it. No further questions. All
right. And we're also uh just in time as I see it. Thanks everyone for attending and for your questions. It was great to discuss with you. And um yeah, have a good day and enjoy the rest of the sessions.
Thanks Yasmin. Have a good one.