PegaWorld 2026: Orchestrating the Next Best Trip: How Transavia Achieved Uplift and Happier Customers

Hello, everyone. Good afternoon. Welcome to the quiet part. Your quiet copilot at 35,000ft. How Transavia uses Pega Customer Decision Hub to personalize at scale. So my name is Joe Alan. I work in the 1 to 1 customer engagement business unit at Pega. But as an ex-client of Pega Customer Decision Hub myself, it is my great honor to introduce this session from another hugely successful client. Joining us today is Max Meijer and Bram van Essen, who bring between them a powerful combination of product leadership and deep decisioning expertise. So Max is the product owner of for Marketing Automation Automation at Transavia, where he leads the strategy and roadmap for AI driven decisioning and real time personalization across web app and CRM, transforming the customer behavior into scalable capabilities that drive measurable business outcomes. And a fun fact for you. This is cool, right? Max actually passed the exam, right?

To, uh, train to be a pilot for Transavia. And he chose decisioning over flying airplanes, right. So I mean, how cool is decisioning, right? Um, Bram brings a strong a strong background as a decisioning consultant focusing on helping organizations enhance customer relevance through predictive modeling, marketing automation and multi-channel decisioning platforms. And he's worked across a wide range of industries. So he's going to bring you a really practical perspective on how to make decisioning work in real world environments. So together, they've played a hugely pivotal role in Transavia journey, from traditional siloed campaign marketing to a far more intelligent, connected and real time approach to customer engagement, and they're going to walk you through that journey today. You're not going to hear just how they did it, but also some of the learnings along the way. And they're also going to touch on, um, the future, the agentic future.

Now if you have questions, please hold those to the end. There'll be time for questions at the end. And with that, I'm going to pass you over to Max and Bram for yours. Cool. All right. Let's kick it off first with a short introduction into Transavia. I think for some people not so well known in this audience, but we are an European low cost airline and part of the Air France KLM group. Maybe a bigger name and well known on this side. Um. Something on the screen. Um, and then our main commercial challenge is mainly on the leisure side. So leisure customers, they only travel with us for 1 or 2 times a year. So catching that customer to book your next trip on our airline. Yeah, it's quite a short timeline to convince a customer. So, uh, an interesting challenge on our side to catch that customer.

And then our core belief. So we're not mainly focusing on, on price, but really on relevance. So showing only relevant offers and not purely sales offers. And then certainly focus also on, on service and even mainly service next to sales. So that's a different approach than, than some of the other airlines. Uh, some quick numbers on our side. Over 25 million passengers a year that we moved towards different countries, over 400 routes and over 40 countries. Uh, I checked the 40 countries, I wasn't sure, but indeed, it's even 44 countries that we have routes to, uh, within, within our our airline. Uh, as I said before, 1 to 2 bookings on average per per customer. Uh, yeah. So that's a bit different than just buying something at a grocery store or buying something at a, at Amazon. So indeed, a short time to convince a customer to, to book their next trip on our airplanes for channels, web app, email and push the order that we implemented was first email, then web and app and eventually then added push.

Um, y CDH and y. We made the move to be more personalized. Uh, first one is the customer journey was getting shorter. Uh, as I mentioned, 1 or 2 bookings a year. So, uh, a very short time to really convince a customer to be a customer that year. Also, then the latest context of a customer is the very most important one. Only booking history is not relevant. I think you have the same example when you book a trip. First it is with your family and the next one is with your friends. Going on a golf trip somewhere in Spain. So the context is completely different. So the latest context really makes a difference in recommending the right flight and the right ancillaries. Another concrete example is descriptions. Of course descriptions are clear, but having that description at our end and acting based on that description is very powerful.

So not trying to upsell when someone had a big delay or cancellation. Yeah, I think that's a well known example of something you just want to avoid. Um, of course, we already were customer of CDH before Covid hit us, but Covid gave us a more a nice reset moment to really evaluate our whole platform, our whole landscape around CDH. So we already had CDH implemented, but it wasn't ideal. We were missing certain data aspects or data points that were really important to really personalize on a, on a higher level. So that time gave us a really nice opportunity to really dive in and have a new landscape that really could help us move to the next step of 1 to 1 personalization and really empower CDH. So let's dive into what we have today and how we are. How we the steps were to get there.

So the first one, I think a big one first, we had quite silent campaigns. And we weren't, we're not thinking in, in, in journey way of looking to our customer. The other one, the big topic is the consistency across our channels. They were all just one separate from each other and they were talking to each other. So offering seats on the other one, offering seats again and again and again. And that wasn't really efficient. And we didn't. We saw that in conversion rates. Later on. I will show some business impact on making these changes. I'll hand it over to Bram for the next part. Thank you. Okay. So we make our decisions basically on three different levels. And the most important one of those three is the booking and post booking level where we do, I think 80% of our efforts is in that level.

Um, the next phase is when we have a customer that we know is coming to our website, we recognize that customer, we know he doesn't have a booking, so we try to get him to rebook with us. And finally, there's also people coming to the website that we can't recognize either because they delete their cookies or they haven't been with us. But we do try to get them to become a customer, to give them recommendations based on prior searches or something like that. So as Max was telling us during the Covid period, we had about five, five, five months of time to, to really redesign our journey. And we really took time to, to get the whole journey properly designed. And that turned out to be quite an extensive one. So let me show you just how long it is. This is our complete journey.

Lots of touch points in there. I won't go through all of them, but it is quite a lot. And our main challenge was to get all these touch points running through CDH. Before this, a lot of touch points weren't even existing or they were in separate systems. So the challenge was to, to, to use this journey and try to plot as many touch points in CDH. Now I'm going to show you some of them. I won't go through all of them, but um, when you um, see this. So first we try to get people to, to book with us. So this is the on the left side on the top, you see a banner, uh, where we try to convince people to buy the next trip. And in this case, apparently Paris was the most likely destination for this customer. So we show our home page banner trying to get them to book.

Uh, below you see, um, A version where we know the customer has searched primarily a certain trip, so we can really make a detailed suggestion with with a known date to it. And finally, on the right is during the booking process. We have certain bundles and we make a recommendation. What we think from a Transavia side would be the best bundle for a customer to book. And obviously, we try to offer a little bit more exclusive bundle than they might otherwise have chosen. Um, like we said, a big part of our business model is trying to upsell existing customers and that we try to do, uh, through multiple channels. So if we, if you come to our website and we recognize you as a customer and we see that you have a booking upcoming in the next two weeks. We try to use that interaction to sell ancillary offers to you.

So on the left side, you see an element that's showing your next flight. You can manage your booking there. And on the right you see more. I'm going to say the offers that the top three offers that were most likely bought by the customer completely on the right side. Um, you see our, our confirmation email. For email, we, we try to use bundling as much as we can. So we try to use a trigger, which is mostly servers or mandatory. Um, and use that trigger to, to include sales and service offers. So this confirmation email, obviously everybody's going to get it, but if we are going to send you an email anyway, we try to include as much relevant service or sales offers as we can. And finally, we have some partner offers also that we try to sell you as much as we can and what we, what we found in that analysis that we did during Covid is that people tend to buy third party offers, partner offers more during closely after the booking, whereas first party ancillaries are bought more closely towards the departure.

So in our communication, we try to do lots of partner offers closely after the booking. So here you see offers that were done on the confirmation page. After the booking is made, you come to a page where your booking is shown. There we can show you third party offers and we can have multiple variations of the offer depending on how many people you're flying with. Where are you flying to? So how luxury of a car you want to get. If you're flying with two people, we might maybe offer a smaller car. If you're flying with eight people, we might offer you a van and stuff like that. And then for the results, I give the word back to Max. Yes, of course, the important part, the results, um, sharing those results within our organization, but also external is always nice, nice to do and keep them involved into, uh, why we need CDH and to improve further within CDH.

So let's share some, some insights. Uh, so what we see, um, on the booking side, so when a customer made a booking, we see 24% increase in ancillary sales per customer, uh, just by using personalized touchpoints within, in those channels. Uh, this is mainly focused on, uh, on email because we have a lot of data based on that, that channel. But for the other channels, we also see quite an increase in in revenue. Another important one is the NPS score for these touchpoints that are personalized. We saw 5 to 10% increase depending on different touch points. So that's also a nice indicator to see that we do or that we show the right and relevant offers. Overall click through rate. It's not revenue based, but it's shows that there's a good interaction with our offers, and it helps the models to learn from those interactions.

So very good to see those high numbers. And then the last one, the one 1.2% increase in returning customer rate, that seems like a small number. But if you see the total amount of passengers per year, it's quite an impactful, impactful number. So the things we've learned in the past six years when making those transitions are mainly focused on the journey part and not purely the data model. So looking towards the customers and the interactions that we see and really talking to customers to see what they need at that time. And that gave us the proper insights to really optimize and get the best results. The second one is timing beats targeting. Of course, we use propensity, but having the right context to have that together with the right offer is really crucial and makes makes a real difference. Um, next one showing results gets and keeps involvement.

Uh, again, what I showed earlier results that we see and get within CDH. We keep sharing those with our stakeholders just to get them involved and get them in as ambassadors for using CDH, and to be able to continue on using CDH within our organization and getting the resources for, for building new stuff. Uh, fourth one, uh, I think it's mentioned before, uh, getting the data right. Um, and I'm not completely agreeing in terms of, uh, getting the data right before you personalize, but I think you can approach it also by, for every use case, if you have it ready. And is it there for a specific use case? Just start with that use case and then build on from that part. Uh, and then you have a solid foundation to extend with more and more use cases with more and more data. Uh, last one, I think that's a real powerful one and really sets us, sets us apart from other competitors.

I always give the Ryanair example. Then you get blasted with all kinds of offers. In many ways we try to hold back sometimes or only focus on service. I think that's really powerful and says something how we see CDH and how we want to interact with our customers. So not always push for sales and not over push and CDH CDH helps with balancing that act. Sometimes we increase it a little, sometimes we decrease it a little if we have some negative feedback. But it's really powerful that you can do that with one system over all those channels. So what's next? Of course, the agentic parts are mentioned in multiple ways in multiple presentations. So also on our side. So let's share the insights that we see on our side happening and how we act on those insights. So what we see on our side in terms of travel, that people are moving towards AI assistance just to discover their next trip or prepare for their next trip.

And that's completely new, at least to us. We don't have control over what is being answered or what's being showed up. So that's something we are looking into to get more control over what's happening in those channels and what we see on our side. We see an increase in traffic coming from those sources. And also the conversion rate from those channels is also quite high. And what we expect for 2020, 2027 is that we see 10 to 15% coming from those channels. Uh, from my side, that's quite huge. So, um, in terms of impact and risk, if you think of traffic moving to the extent of 10 to 15%, for us, that could be a traffic and also conversion within 6 to 800 million of revenue shift in terms of channels. So that's quite big. So then we better need to be on top of those channels and have some control over those channels.

So that's our aim for this year but also for next year. First of all, we focus on exposing real time data in that channel so that we at least show the right information so that the flight is available and that it has the proper information of availability of ancillaries and other stuff. So that's our first step. And then the next step, I think also where CDH comes in and that we want to find out also here in PegaWorld, how can we control and recommend based on that availability? We also this morning also already in the CDH breakfast. How can we control or recommend, and what is the role of CDH in that? That part. So very interesting to learn more about that, how that could be part of what we're already setting up. So really looking forward to diving in on that topic. I think that's it on our side.

There are some questions from your side. All right. Hello. So you had some impressive KPIs. I want to say about five or so slides back over the last 5 to 6 years. I'm sure that that wasn't the time frame it took to capture value. So I'm wondering what that journey looks like in terms of the overall impact of having Pega in place and that story that you can maybe share. And then in terms of impact, you mean in transformation or the shift from 5% to 25%? I think the shift, right. So you shared numbers from a five year journey. I'm wondering what it looked like one year in or two years in. Um, if you can share, if you have any insight to that, that would be interesting. Yeah. Of course. It's, it's a, it's a learning process, or at least it's a learning phase.

It's the same for models. Models need to learn. And we also need to learn what to put into the models. So it's not that we straight from the bat got those numbers. So that's something that we're doing currently as well. Looking into the predictors that we have within the models and seeing interactions and keep keep an eye on those those numbers if they increase or decrease. Um, yeah. So that will be my lesson to someone who just started also diving into the modeling part and see what happens there and keep optimizing or adding predictors to be able to see if they have impact in the outcome that we have with our customers. Yeah. And I think on a on a use case level, you see quite vast impact. So on an overall level, this this is not done in a, in a year or something, but on a specific use case level, you see the impact almost right away.

So when you start personalizing and using models, the impact is quite quickly. So so use case wise, you see quickly results on global on the whole journey. That takes time. Can you share what you mean by quickly. What are you talking days? Weeks. Yeah. So depending obviously a little bit on the use case. How how many responses are you going to get. So it takes time to learn the model. But but some users use cases that we have. We saw results even after 1 or 2 weeks. So it can go really quickly. But obviously it's depending on on on. The volumes. The volume of your of your of your use case. So I have a question in terms of the communications that you've built out for this engagement, are you using Pega to build the emails or do you have another tool where you're, let's say emails are built and how are you doing that?

Yeah. It depends on the channel. But for email, we have an external email service provider. But the complete like context content of the email is coming or content we send adjacent to the email service provider and then the emails being created in this email service provider. Got it. And then I think you said you were able to really expand your consumer journey. Um, there were lots of gaps and you added more on. So just from an overall scale perspective, can you say something about how many emails, let's say you were sending before this transformation? And then after from a consumer outreach standpoint. Before, maybe before. So there was an increase in terms of amount of amount of emails, but it's mainly about the NBA that we put in the email. So I think we added some touch points, uh, but we mainly, uh, transformed the existing touch points that we have and included more MBAs into those touch points, you know, and it's still a balancing act, right?

You don't want to overload the customer with emails, even if it's in a short period of time. So, uh, it's a balancing act and depends on. And in some cases, it was just, we already were sending the emails, but we were sending them not personalized via a different system. So we tried to do it via CDH. Yeah. So, so in terms of number of emails, I'd say the same, but it was just more personal and, uh. Yeah. And one thing. Yeah, it depends on how far before the departure someone booked. You have a shorter time frame to communicate. So that also makes a difference in terms of how much you send within a certain time period. Yeah. Hello. My name is Rakesh. I'm from Citizens Bank. So in a very different industry than you guys. But then um, you guys talked about CDH journey. So one thing that stuck with me is timing beats.

Um, um, something along those lines that you had. So the targeting. Yeah. So the essential thing, what you are trying to say is that, you know, aside from the propensities, you need to focus on which part of the journey the customer is currently in so that you are targeting them with the right kind of messages. If I got it right, right. Um, now with CDH right, a customer can be part of multiple journeys at the same time. So my question around that is if not propensity, how do you prioritize? Right. Like I can just think about something like from like, if somebody has booked a ticket, right? You might be able to sell them a travel insurance, or maybe you might be able to sell them an upgrade. So how do you basically align that targeting also to be based on certain, uh, certain order or ranking?

Right. And how do you ensure you put the customer into the right, um, journey amongst all that they are eligible for? Thank you. Yeah. Yeah. So, um, so we're using the arbitration formula of Pega mostly for it. So, so we use propensity obviously. And we have some some value also added in there. So we try to, to use the, the outcome of the formula to, to show what we want to send. But it's also based on a business ruling. So so some emails are sent five days before departure. Some for instance, um, when we, uh, well, we were one of the last airline companies that were charging people for their trolleys. Uh, but when we did that, a lot of people still took them and we had to, we had lots of stuff at the gates because people had to pay for it. And that was not very nice.

Uh, so that was for us a signal. Okay. We need to communicate more direct to people that they have to pay for their hand luggage. If they don't have, if they didn't buy it. Uh, so, so we inserted a touch point in our journey saying, okay, three days before somebody's going, we sent an email saying, uh, attention, you haven't bought cabin luggage, so be aware that you have to wait. But we try to use that touch point also to maybe sell them hand luggage if they hadn't buy it. But it is a service communication. So um, yeah, we try to use business logic as well as arbitration to form the ideal message. So is it like always a service triumphs, uh, upsell or something? Yeah. So we try to use as much the service, the service moments. We try to use those as much because they, they have to be sent anyway.

So if you're going to send an email, for instance, confirmation email, you have to send that internet check in email, you have to send it. So we try to use those moments also for for commercial point of view. Got it. Thank you. Thank you so much. So, um, my name is Nagesh and I am from the same bank. I really like that slide the five things. Slide what you have. I was more interested in what you said. Restraint is a feature. So does that mean how I'm interpreting is you had a, let's say a customer came on and there is an opportunity to present some message, but you decided not to even show them anything. How is that? How does that translate that? Yeah, I think it's more in the contact policies so that we at least have a central place to control the amounts of contact we have as a specific customer.

So it's always balanced and we play around with the amount of times you could communicate with the customer, but have control over that across channels is quite powerful. And that gives you the opportunity to also remove sales offers. For example, if there's a disruption or increase, if there is a campaign or a there's a gap of flights we need to fill. So you're quite flexible in terms of offering but also suppressing specific customers based on certain ruling that we have in place. And so also we we, when you return from a flight, we usually send you a welcome home message. And when we know that you have a big disruption or a big delay, we change the content of the email. So we still send you the email, but it has a different content and a different tone of voice depending dependent on how big the disruption was.

Got it, got it. Makes sense. Thank you. Hi, I'm Ellen from the Lcbo in Toronto in Canada. Are you using your MPs data point only to suppress or any journeys against like say, negative MPs? Uh, no, it's not used as a value to show or not show specific offers, but it's used to report on. Just an outcome. Yeah. Performance of or an indication of are we doing the right thing in the right touch point? And it's just a score that we also report on to, to the board how we perform overall. Yeah. Okay. Any more questions? Thank you. Hey, thanks for the presentation. Um, I'm curious about, um, maybe going back to the point on propensity models, um, have you used those capabilities to, um, try something on pricing, like doing tests and like changing the, the, the price of products that you're selling or maybe doing some variations at a customer level, thinking about a B testing and things like that.

Uh, yeah. We are not allowed to adjust pricing for specific customers. So it's always generic in terms of we can do it based on availability, but not for a specific customer. We We cannot change price or show different price points for different customers. So it's not really connected in terms of pricing. Models are being used to set the price, but that's a different department. Yeah. But it's not based. On and it's based for everybody. So it's not on the customer level. But it's not connecting to CDH. OK. Anymore. Anybody else. Any questions? No. Final call. Final call for questions for Transavia. No. In which case, thank you so much to both of you for a great presentation. And there'll be around. If you've got any more questions you want to ask, I'm sure they'll come to the side of the stage and you can ask them privately if you like, but thank you very much.

Give them a round of applause.