This interview is part of our exclusive ‘Airline Leaders Series,’ conducted in collaboration with Branchspace – digital reinvented, aiming to transform airlines into better digital retailers. Throughout the series, we’ll highlight the key concepts that are shaping modern airline digital retailing, ensuring you gain valuable insights from each discussion.
Leader: Liliana García Magaña
Airline: VivaAerobus
Role:
Location: Amsterdam, Mexico City
Talks about: Data, Web Analytics, Customer Data Platforms, Artificial Intelligence, Machine Learning, Digital Experience, Data Literacy
Why do you need to read this interview:
Liliana is a standout leader with a strong background in data, analytics, and business intelligence. Her expertise goes beyond what is commonly found in data and analytics leaders, showcasing an in-depth understanding of data-driven decision-making, detailed implementation of airline digital analytics and tracking, and the pivotal role of data in enhancing airline digital retailing and digital marketing strategies. With a pragmatic and straightforward approach to AI and machine learning, Liliana brings a fresh perspective to the world of airline commerce. In this interview, she methodically unveils her approach to harnessing data and AI for the benefit of airline digital marketing strategies, providing a valuable blueprint for any digital leader seeking to excel in the data-driven landscape of the airline industry.
Liliana: Addressing Viva’s state and goal for 2024, it’s going to be certainly a challenging year. Of course, probably everyone’s heard about the issues with the Pratt & Whitney engines, so that means a lot of aircraft on ground, which puts pressure on the business. So on that end it’s certainly going to be – there was all this organic growth coming. We have to control that growth because there’s not enough aircraft, which is a situation that not only affects us but many airlines.
However, that also gives opportunity for data to become a star. That’s an input that we already have. So 2024 is going to be really about, we have all this data; how do we make the most out of it? That translates into some very specific objectives. First is a big leap into machine learning. I think now, at least in Viva’s environment, now that we have been building, all the reporting infrastructure, cloud architecture, making sure that we have really good foundations – 2024 was always our year to take that big step into machine learning with the use of internal tools or the actual tech stack we have. We have a roadmap of all of these small initiatives, some might not be so small, but that’s definitely one of the big turning points.
That said, the next point is data maturity. We have so much data we produce not only in our reservation systems, but we have a very robust model in our web tracking, both for website and app. We really need to make sure that all of this data is used in the best way possible. Evolving the digital marketing maturity, using all of this data is key for us. One of the big initiatives we have to enable this is implementation of a customer data platform. We have spent all of this year looking into platforms, going into the depths of if they can actually pull identity resolution or not.
An issue within the airline industry is that it is very difficult to have one single customer profile, and within that, all of the data that comes along. The next year, we are going into that direction – after doing all these evolutions technical and business-wise, we’re moving towards the implementation phase and then making sure that this data itself and then all of the segmentations and the machine learning models can be pulled on top of that data and customer profiles can be used by the marketing teams.
Liliana: The last four years, we have been building the foundations, including the web analytics strategy. This year, our focus was to really understand the technology that we needed. We have all of that data ready, in the cloud, and ready for use, but now we need a platform that allows us to unify the customer profile and activate it very easily. Our approach in this case was, let’s take the time that we need to make the best decision so we make sure that our investment is the best and it also actually enables the user identity in the different ways we have it. Like a user is logged in, not logged in, and so on.
So this year we were fully into evaluation mode and proof of concepts and understanding the depths of the CDP’s implications of cost and the different ways of activation and the operational model that we want for next year so that we can actually move into the implementation itself. So this year was really a lot of planning to get there.
There is one more. It’s the leap to AI.That’s probably an agenda for many companies, many airlines. I feel many of them are already doing it, let’s say the bigger ones. But I think the objective for us next year is to identify these low-hanging fruits and identify which is the best way to enable these low-hanging fruits, whether it’s for operations or for revenue management or for personalization, like prediction and so on, and to identify this roadmap to get there.
That involves also understanding the level of knowledge and education that all of our teams have in AI. Part of that big leap to AI is also making sure that our leadership has enough knowledge to make a decision about investing in AI. And then of course implementing some of these. One of those low-hanging fruits can be, how can we nail data democracy and easier access to generative AI? How can we make this faster? Or there are some other topics, like how can we use generative AI to create all of the marketing creatives fasterand in a more scalable way?
That’s one of the biggest initiatives for next year. We really want to move in that direction, but we also don’t want to move there so in a rush. Many companies are doing it and airlines tend to just run towards AI without understanding the depths of AI and the actual possibilities and implications. So, we’re really trying to create awareness internally. Like, we are getting there, but let’s do it in the best way possible.
Liliana: I spotted a few already. Most of them revolve around AI and all the different components within AI. I see several startups driving generative AI initiatives and addressing customers. They are trying to get customers to start growing into that space. I think that has a good and a bad side. On the good end, you can get a lot of proof of concepts going early for a lower cost. On the other side is that with startups, there’s always a risk. I’ve seen plenty of startups reaching out to companies to help them design these tools.
There are some other more data-specific trends, like BI/AI driven solutions, like Tableau Pulse, for example. Also, with the current tech stack that you already have, how can you leverage large language models and so on within that infrastructure? I think that is one of the ones we really want to explore next year.
The next trend I’ve seen is, again, on the generative AI, but more for image generation. There are many companies that were doing web analytics; now they are also evolving into, “Now, because of the tech stack, I can do a generative AI for image generation for marketing or email” and so on. So that’s another thing I’ve spotted. We have actually had a few companies reaching out on that end.
One of the emerging technologies that I’m actually curious if it will get there at the same time is more in the data science / machine learning space, like DataRobot, for example. I’m just curious about how these companies having focus now on machine learning, let’s say low code machine learning, and deployment, how they will actually compete with gen AI but also – not open sources, but tech stack like Google Cloud with Vertex AI and so on, so you can do your own models. I’m curious about that trend.
But I think generally, everything revolves around AI. And some of the tech stack you already have, even CDPs or personalization tools, they are all adding AI components, whether it’s actually machine learning or whether it’s more like large language models and so on. They are all trying to adapt what they have with AI.
Source: VivaAerobus
Liliana: Definitely one of the challenges is, depending on your business model, if you’re an ultra-low cost carrier, everything goes to the bottom line. Investment has to be very, very specific and very well thought out, and it must have a return on investment.
Sometimes the benefits of these AI initiatives are uncertain until you actually know. So financial evaluation of AI, particularly within ultra-low cost carriers, can be a little bit harder because you also want to keep costs low, and by default AI initiatives will have higher cost maybe than other initiatives.
That leads me to the next challenge, which is we need to really train and educate staff on AI theory, deployment, for actual costs for investment. What’s the cost for deploying or building machine learning versus computer vision? Both have totally different costs. That will be really tied up with the actual financial evaluation of the investment for AI initiatives, and we need a good assessment. So we also need to make sure that all our executives have enough knowledge of that.
I’ve read many articles on companies, and when they quickly move into AI, it seems they are forgetting the baseline. It’s like, okay, AI is a big buzzword that everybody thinks they understand, but few people actually fully know in depth what it means and what it takes to get there. So we need to make sure that that foundation is also set in place.
Also, how can we leverage our current teams to train them in more specific skills – say like machine learning. Many of our team started as data analysts, data engineers, web analysts, and so on. So now, we have to set up training for them so they can also develop these skills. You’re not just going to re-hire your full team with machine learning engineers because that’s not going to work either. We have to create this path on building the foundation for them and how to leverage our own current tech stack.
I think also finding partnerships is one of our main opportunities in general. When we’re looking also for partnerships, say with Google for example – all our tech stack is based on Google Cloud, so we are also trying to create a stronger relationship, a partnership with Google on how we can use their products and all the components within Google Cloud to enable our own initiatives, because that’s a tech stack we already have. That doesn’t need any additional investment other than the billing cost of the use of the tool. So we’re also trying to create partnerships with them to develop these.
Liliana: First, I don’t see that many changes in headcount, particularly because of the constraints that we have next year in terms of growth with what’s happening with the problem with the engines and the financial impact it has on the company. The challenge then is we need to make sure that we can enable our own team to create these types of initiatives. There needs to be a lot of internal work, and reshaping some of the positions that we currently have. More than additional headcount, it’s reshaping positions and evolving them from data analysts. Like changing some data analysts into data analysts and machine learning specialists.
In the end, we don’t know everything. There are a few people in the company that know in depth about AI. So we do need to find these other companies that can work along with us to build these capabilities. We are actually looking into the startup approach, seeing which startups we can find win-win situations with. In terms of hiring just internal resources, we’ll probably outsource and work together with consultancies or startups to build all of these AI products.
Liliana: We are looking into implementing a customer data platform. We are also doing evaluation of a new personalization tool. Our focus now has been on also understanding what type of tool will enable us to be self-sufficient in enabling these personalization tools. I think there’s a set of small initiatives on things we can actually do in personalization, but not many platforms can actually pull that off. Many of them just go into the space of pop-ups or so on. But we want to go in depth and push for dynamic funnels, or dynamic sorting of ancillaries, for example.
Being a low cost carrier, it’s also sometimes difficult because you don’t have the flexibility of personal discounts and so on that many personalization tools can enable. So we have to figure out how we can embed personalization in our current digital flows to increase conversion. I think those quick wins are what we’re focused on. Like, how do we make sure that if Iztok is flying with a length of stay of 10 days, the first thing he can see is an extra baggage of 30 kilos because you’re staying longer, and then you have a personalized message of “Hey, we recommend that you add this suitcase now; it’s cheaper than at the airport, and you’re also staying for 10 days.” Then we’re not changing any pricing or anything. We’re just giving the customer the right message at the right moment.
At Viva, the data team is relatively young. We have been here for almost only four years. Compared to other airlines, they have had a data team for longer. We really took the time this year to think in depth of the strategy not only in terms of technology but in what we actually want to achieve with that and what we need to get there and how we’re going to operate it once we have it.
Once we have those points really structured and aligned internally and connected, then we are in a very good place for starting next year. I’ve spoken with some peers from other airlines, and their martech is not always connected. Sometimes the CDP is in silo, the personalization tool is in silo, then email marketing, then web analytics. They all form different teams in different buckets. So I think something we have done right is that we have it all connected and under one team. We are making sure that everything in the web analytics model can be leveraged with the CDP, and it can also be leveraged within the personalization platform, and that all those audiences created in the CDP can be straightaway activated in the personalization platform.
And if we need to personalize something, we can also connect it or A/B test part of the capabilities that we are looking to see with personalization and we can actually look into the email marketing campaigns. So we’re really making sure all of this is telling the same story, along with the customer relationship management system that we have for all the complaints and stuff. It took a while for us to put that strength together, but we are in a really good place because now we are making sure every tech stack speaks to each other, and there are no silos.
And of course, with the risks this can have in terms of customer experience. If they don’t connect, first, you don’t monetize enough, but then you also have a risk of deploying experiences that are not in the best interest of the customer.
Liliana: If we think about the industry, what I see happening – and I’m not talking about Viva itself, but just in general in the industry – a lot of AI travel assistance. I see that happening – I mean, of course, ChatGPT has deployed with Expedia where you can find easily your flights, hotels and so on. I see that happening a lot next year. In my honest opinion, I’m not sure actually how much that will generate. I think it will be more of a fancy nice-to-have than will actually generate revenue. But AI travel assistance, that’s for sure. I see also a new way of chatbots.
Prior generation of chatbots were rule-based. They were like “if-then, if-then, if-then.” I think next year there should be a big – and I think we are there already in general. We should be there, enabling or building these chatbots based on our actual data and our customer policies. All of these different questions that a customer has when they go to the airport or in any part of their customer journey. Sometimes there’s people that don’t even know which gate they have to go to, or that they have to go to a gate, if they’re first-time travelers. So an actual good AI chatbot should be something that next year is really covered.
And probably last is more dynamic pricing through AI, both for ancillaries and for the actual fee of the trip. I think most of the airlines have legacy systems, and there have been many companies trying to do dynamic pricing, but they haven’t really gotten there. Because of course, it takes a lot of integration to have this fully automated. So I’m not sure if this will actually happen next year, but that should certainly happen in the next couple years: you have a fully automated dynamic pricing AI model that can connect to your own website and the reservation system, because the two interfaces need to be on the same page. Then they can actually manage to get that dynamic price to the front page of your website or your app.
If you want to learn from leaders like Liliana about how to advance your airline digital retailing or want to be the first to know when our next Airline Digital Talk will be published, please:
No Comments