Christian Rudolph, Lisa Dahms, and Philipp Böschen

Hello, and welcome to our talk for the DevOps Enterprise Summit 2021.

My name is Christian Rudolph. I'm working at the TUI Group as the Head of DevOps Transformation, and I have today with me Lisa.

Lisa Dahms: Hi, everyone. I'm a Senior Learning and Development Manager within Group IT.

Christian Rudolph: And Phil.

Philipp Böschen: And I'm a DevOps Evangelist and Coach, helping other teams develop in that space.

Christian Rudolph: Let me first introduce you to TUI Group and who we are and what we are doing.

The TUI Group is the world market leader in tourism. Before the COVID crisis, we had more than EUR19 billion of revenue. We had more than 27 million customers we delivered travel for, with hospitality and markets as well. We were a well-being company, we had a good growth estimate for the next financial year, and we already had more than a 17% increase of bookings in January 2020 compared to the previous year.

How are we organized? TUI as a company is very well spread across Europe. We have around 15 different countries that we operate in as a tour operator and sell holidays, travel, and hospitality to people in the countries, but also in some other areas where we have only online travel agencies where we sell our own holidays to different destinations across the world.

We have the 15 countries divided in three different regions. We have a central region, which consists more of the central European countries. We have the northern region with all the more Nordic-oriented countries. And we have a western region, where we have markets like France, Belgium, Netherlands, and Luxembourg.

Overall, we were planning to take a look at our competition and also our new disruptors in the markets like Airbnb, TripAdvisor, and Google themselves, which brought a lot of digital disruption to the travel and hospitality markets, especially in the tour operator business. How do we book our holidays, what are we doing, and how do we make sure that the digital journey for the customer is present to them and helps new customers get the right destinations and the right travel environment for them?

We also had to look at new innovations. There was a lot of new technology out there, like Google Glass, where you can have augmented reality inside the destination to get insights about cathedrals or museums, to give you a good opportunity to show what is possible in a new area of travel and how do we lead that. One of the main points TUI overall stands for was to help lead in this area. So we visited a lot of the new technology companies. We worked with a lot of startups to embrace new things in the travel industry.

But everything was more led to a long-term plan. Yes, we did need to change at the moment as well, but this was more smaller changes and not a big disruption as we maybe have seen now.

How did we plan to do it? Overall, TUI had board meetings around how people would like to transform. The buy-in from the company was very well there. We understood through the digital players that we needed to do business harmonization across the European markets together with application landscape consolidation as well.

And we already recognized, okay, we need to change the skills in our workforce. We needed to upskill certain people. We needed to develop new skills which are needed on the market. We then started to work with press statements inside our company to make sure that everybody understands where we would like to be heading. But everything was still on a more long-term plan, about three to four years at minimum, to get to this position and to make us successful afterwards.

Skills shortage, we very well discovered that we know there's a lot of things which will happen, and there are a lot of studies out there about how the skills shortage will hit our industry. It's not only our company; it's the whole industry that has specific skills gaps in the AI, data analytics, and DevOps space for all the demand which is coming to the different areas of the industries and also other companies.

One of the things we envisioned already during this period of long-term planning is that we need to invest in our people and upskill them. We started with a big cloud program already at that time and now thought, okay, how do we bring this further? How can we bring it closer to where we have the needs for the different people, but also take all the people with us?

And then COVID came around.

To give you some numbers about what this means for us from an impact perspective: we had 91% reduction in customers in the last financial year. We lost 98% of our turnover. And we had to apply for financial aid from the German government and overall from the different investors. So overall, it was a financial package around EUR6 billion, which we now carry with us in some areas.

This was tremendous. There was a time in March where we had to think about what this means for us as a company. Very fast, the board came together. A lot of things stopped, and we had to work together: what do we do?

One of the targets was, can we reduce our cash out? So we reduced the cash out by 70% to make sure that we don't spend money when we don't get new money in.

We also started to think about our assets. Our company has a lot of assets in the hotel area, in the cruise ship area, in the aircraft area. What can we do to make this a benefit for us from a financial perspective, with the different models which we explored there, and to make it visible for us?

And we also had to reduce net burn rate. Everything that is cash out from the company had to be revisited and re-looked against that.

What action happened then for us? From an IT perspective, where we will mainly focus today, we lost one third of our workforce in IT. This was mainly contractors we had hired to help us do certain things, and we reduced our contractors from more than 300 people down to 50.

We also started to use benefits of the furlough schemes in the different countries. In Germany, we call it Kurzarbeit, to make this happen where you get a certain amount paid from the government, to reduce our IT capacity. Overall, we reduced the IT capacity by 55% of people's work there, and we stopped nearly everything around that.

All HR development activities were paused, and we also went tremendously from 330 to 45 projects which helped to survive the business. There were only 45 projects which had to continue. And what does it mean in detail? Phil will now explain to you.

Philipp Böschen

Thanks, Christian. We've seen a general overview of how the TUI company had to cope with the pandemic and what that meant for us overall. With this, now we're going to dive a bit deeper into the IT organization.

We've just seen that our workforce has reduced drastically, so we had to think about how do we use the existing leftover people, which we still have time with and for, to work smarter, not harder. How do we actually leverage them better than we used to do? We can't just throw resources at problems anymore.

Also with that new reality, like this pandemic, suddenly we had to cope with lots of different little challenges, like bringing people back from their destinations where they were still stuck, so we had to cope with developing new features to get them into their new homes and their destinations.

That whole reduction in workforce also meant that everybody had to take up new responsibilities. Everybody had to look left and right and help out. We had to do less shoehorning into specific roles, which then also means the skill landscape for our workforce has changed a lot.

And all that still happens with 98% loss in turnover, so we cannot really go out and spend new money. We also have to always be very smart and very due diligent in how many resources and when we're going to go into which resources.

We now look at the four areas we as an IT organization focused on. First, we're going to look at DevOps coaching. How do we get the learnings from DevOps, things like DORA, things like the framework from McKinsey, how do we apply those to our organization to then help bring other teams along? That's why we have some central experts who coach and then go further.

Then we go into metrics, which underline the entire thing. Without metrics, we can't track anything, we can't prove anything. That's where we go and lay the foundation to then support the organizational change.

Christian already alluded to an ongoing change, but we had to accelerate that and move along and make that a bit quicker and change a few other things. Lastly, we're going to move into an area around a global upskilling program led by HR and driven by them to help us develop the IT workforce into this new landscape, this new skill variety.

If we look first at the coaching, what we did is we took enthusiastic members of the overall developer community and built a thing called the Community of Practice for DevOps, which was at first a grassroots initiative of just like-minded people thinking about what can we do outside of the DevOps toolkit to make things better, to make things more efficient, to help people along.

You can see at the bottom that we had a few presentations around what you can do. The left side shows the metrics bit: where do we lay the foundations, what metrics do we track? The right side shows the middle, where we developed a maturity model, which is a self-assessment for a team to go through and ask, okay, how mature am I in this space? It tackled agile, build, deployment, security, and other miscellaneous things we just find interesting.

Additionally, we went from the methodology bit to a bit more practical, technical bits that people could apply. We built up a central template library for infrastructure templates and application templates and just went along with those to help people adopt good practices and show them at an example, less at a PowerPoint level.

That's our coaching bit, and this is how we started supporting teams and helped them along to get going quickly.

Then we moved into some metrics, and some of you will probably find these familiar. These are very closely aligned to the DORA metrics. We started first to get the delivery stable, because without a stable foundation, our delivery can't really cope if we put more pressure on it. The more pressure and speed we want to put into it, we have to pay the price of stability.

What we did is we went through our existing version control systems. We tried to get people to harmonize into one, where we put a lot of work into harmonizing people into one version control system, so we then can simultaneously also get the deployment metrics out of it.

You can see a graph above on how we started collecting these. In the red, you can see pipelines where we've learned something, and the green is the successful deployments. You can see that over time, the deployment frequency and the amount of deployments actually changed drastically. Towards the end of this graph, which marks April last year, we can see the amount of deployments has doubled in size to six months prior. Also, the amount of good deployments is growing while the failed deployments are stagnating.

This is where we're currently building a base to help identify teams where the coaches and central enabling organizations can help them along and help them with their struggles. This is one really key thing that we find.

The next thing is the cycle time. Once you've stabilized the delivery, we now need to adapt quickly because we are in a drastically changed environment. Christian alluded to the innovation bits, and the pandemic hasn't made things slower. The overall environment needs us to adapt quite quickly, which of course for an enterprise usually is quite a challenge.

First, we want to establish how fast do we cycle for any features, what is the cycle? We can see some teams started to actually really track that nicely. We can see the curve going, we can see how they deliver on average, and then we can compare them to themselves a few months ago and see what changed. How can we get this going?

The faster the cycle time and the smaller the batches, the better the package gets tested, and we can adapt quickly and throw away work that's wasted because the environment has already changed again. Now regulations change like every two weeks, so we have to be quick on our feet for that.

The last metric we wanted to look at is while going fast, you build up a lot of debt usually. We looked at the concept of technical debt, which is cutting corners while developing stuff, to then see how much debt have we paid. We put a lot of effort into amplifying and showing that metric so we can now tackle that.

That's one of our current things we want to look at: don't just deliver at speed, don't just run in a direction, but make sure you can deliver the quality so it doesn't break two months later. This is one of the key metrics we have right now, which we are looking at and tracking for the teams and helping them. Can we maybe develop this with more quality in mind and less rush?

Sometimes the cycle time reduction actually comes from developing more quality into it. Reducing the tech debt buildup will actually improve your cycle time over time, we find. The less tech debt you build up, the more stable your delivery gets and the more stable your cycle time gets.

With these three metrics in mind, we've then moved on to the organizational bit. What has the IT organization, the organization overall, looked at to get to that point?

We have to limit work in progress, so we get faster feedback on the work. We need to remove the context switching, so the fewer topics overall for the company, the less people need to switch between differentiated goals. If everybody has the same goal, every conversation they have is in the same context of that goal. Overall, this leads to higher velocity. This is from the lean mindset: if we deliver smaller things continuously, we increase velocity.

And then we have shared goals, which moves the same idea. If we share all of our goals commonly, then we have less conflicts and less conflicting goals, and suddenly everybody can see that they're doing their part for themselves.

The next thing is, after refocusing, we can now accelerate. The acceleration is basically where we say, okay, we move to a product-focused organization, where we say, okay, we do it once, but right.

We focus in our culture that we embrace remote working because we've seen amazing benefits from that. We go digital first. We go away from old, big legacy processes, and we want to embrace leaner digital processes, which we've seen work really nicely because we've been forced to adopt them.

We want to embrace an open knowledge-sharing culture. That's why the coaches, for example, openly share all of their information, and the templates are shared, and we try to get teams to talk between each other to help each other out.

You've seen Christian present before where we had regional markets and a group IT, and then every market had their own IT departments, which then did everything slightly differently. We still have three reservation systems. That all moves into a separate organizational structure, which now tries to do things once. So we build one reservation system, we have one sales system, we want one system building products, where we try to leverage this product organizational thinking.

With these organizational changes in mind, the skill landscape changes a lot. To support that on a global level, we've got Lisa with us, who's leading the for:ward program to help us develop our workforce better. At least we're trying.

Lisa Dahms

Christian mentioned in the very beginning that the skill shortage is out there. Everybody knows about it. We had to figure this for our company, and we had to see what can we do in order to avoid that our skills are not future-proof.

Back in 2019, we came up with the idea to develop or set up a development program mainly focused on analytics. Then COVID came around, and we realized we need to have DevOps in the focus of this program as well. That's why we set up a program called for:ward.

The program is really new to the whole of IT, because never before has there been a program mainly focused on the whole of IT. It's a global program. In the past, we had local initiatives in the different training departments, but now we really have a global solution that we can offer to our IT people.

The new thing also is that it's centrally funded, so that makes things easier. The guys were talking about increasing velocity. Centrally funded things make things easier for us to increase learning activities because departments don't have to worry about how to get the budget for it. That makes things easier as well and enables our people much faster to get the skills that they actually need.

In the heart of this program, we have two streams. One piece is role transition, and the other piece is on-demand learning. I will give you some more details to this in a couple of minutes.

Of course, we will have to have our new work mindset in our minds. That means we have to have a safe-to-fail environment. People need to talk about their experiences, their learning experiences much more. We need to encourage people to take the hours to learn and do that during their daily business. That's not a common thing until now, so that's something we have to change.

Of course, peer learning and interaction is one of the key aspects in order to share knowledge across the company. That's something we have to focus on with the program as well. We have a vision, you can see it below here, that's called: at TUI, continuous learning is a natural part of our business delivery. That means we have to become a learning organization, and we're not there yet. That's why we try to implement the program, or we have started to implement the program.

Here are some more details about the for:ward program. I told you that role transition and on-demand learning are the key pieces of the program.

Role transition really focuses on enabling our people to move into completely new roles, for example in the field of data and analytics or DevOps especially. We also know that certain IT roles will be outdated in the future. We try to bring those people, or very good IT people, into the position to learn future-proof skills and make their transition much easier for them.

The key thing here is that those people taking part in the role transition program will get time alongside their job to actually focus on the program, and this will be for about three to six months. We enable those people to reduce their work and get more time to actually learn and transition into their new role. That's really, I think, a new thing, at least for our company, probably also for other companies as well. So the role transition piece is incremental for our journey to the future.

Then we have our on-demand learning piece, which means practically that you can learn any time, any place. Especially when trying to become a learning organization, you need to have those options for the people. Our employees should be able to decide on their own when they have a demand to learn something, and when they have the time to do it, they should be able to do it. This is our chance to help our people to learn whenever they want to and whatever they want to, and to give them more self-responsibility for what they are actually doing.

We have started with our on-demand learning stream with a pilot in March this year, and we are just about to kick off our role transition program in May 2021. Okay. You see, I'm coming from HR; it's not as easy as for the IT guys.

Here are some more quotes from our people. What we hear coming from the organization is that they have just been waiting for a program like for:ward to come forward. That means from the top to the basis of our organization, people are really eager to participate in the program. They tell us, "It's really something we have waited for. It helps us to make the learning easier, to include it in our daily life." That's exactly what we're trying to do. And if it helps us to accelerate our DevOps journey, then it's just the right time to do that.

Concluding everything that Phil and Christian have told you, and that I've been telling you, we figured that crisis is something that we can do. We knew that before COVID came around, but COVID has shown us that we, as a company, have the best people to cope with all the challenges that are out there.

Even with reduced financial background, Christian told you about it, even with all the challenges, skill changes, workforce reduction, and so on and so forth, we still were able to enable our customers to come back from where they were stuck. We enabled them to go on a new journey wherever they want to go in the future. So, yeah, crisis we can do.

Also the coaching aspect and the upskilling pieces: we know already that you have to reinvent yourself over your career probably 10 times because the skills have such a short life cycle. That means we need to focus on sharing knowledge in the company, and that enables us, for example, by bringing coaches into play or by offering learning solutions like we do with the for:ward program.

What also helps us, or what helped us getting through the crisis, and I mean we're not there at the end, but having clear and shared goals is really something that is making us better and focusing. You heard about it. I think whenever there is something that we would like to do, talk about it, communicate about it, and tell the people why it's necessary to do so, and then they will follow you in delivering those goals.

But some challenges remain. We are shortly before getting back to our offices, when COVID enables us to do so. There will be impact on teams as well, as we now see that the cultural changes coming forward from the new domain organization are a challenge for our people. This will be interesting for us to see how this develops over time.

Also, we have to make our technical debt decrease, you heard about it, and continuously measure our practices that we put into place. I think KPIs are necessary to focus for the future and to always reflect on how successful we are with the things that we're doing. And clear shared goals, as you heard about it, are the main important thing.

Now we would love to hear your experiences and feedback. If you would like to speak to us about anything we presented or have a good insight about one of the challenges we will face in the future, speak to us, connect with us on LinkedIn. See you later on the Slack channels, and thank you very much for your attention.

Thank you so much. Bye.