Host Intro (Gene Kim)

So one of my favorite leaders of study has been Fernando Cornago from adidas. In 2017, he and his team from adidas attended this conference, and they left inspired by a specific presentation that Jason Cox gave from Disney, who we heard from yesterday. This actually led to the creation of a centralized technology group to better advance the goals of all the business units through engineering excellence. And so Fernando is now VP of Digital Tech.

Over the past couple of years, Fernando has presented on his ever-increasing responsibilities, and as they kept adding value that people appreciated, it became the primary revenue channel during the pandemic. He's talked about cost control and this massive focus on improving their e-commerce capabilities.

So Fernando's going to share an update on that journey, including some findings from a GenAI pilot they did that included over 500 technologists to explore how it could better enable people to do their work easily.

Here's Fernando.

Fernando Cornago

Hello? Hello, everyone. Yeah, it's working. Hello everyone, and thanks, Gene, for your words. So my name is Fernando Cornago — and sorry, Jason Cox, I have the best job in the world, not you.

I have had it for the last nine years at adidas, where I had the honor to lead the first software engineering team, the first one to scale to seven technology hubs and 2,000 technologists across the globe. I led the platform engineering, our cloud strategy. And currently I'm leading our digital tech ecosystem, which is how we call our consumer-facing stack.

For the last seven to eight years, I try not to miss this event because I really always get more than I give. Today I had to cut a little bit last minute my presentation — thanks, Ann Margaret, I will do anything for you. I'm Spanish, I can talk very, very fast, so I will make it.

I want to use this time to say thanks to this community. I will present six, seven things that I took from this community and show you the results after six years. And then I will give you a glimpse of how GenAI is changing our world in adidas tech.

But first of all, we don't work for the sake of throwing technology to the cloud. We work to change people's lives through sports. It's really our belief. And in order to change as many lives as possible, we want to be the best sports brand in the world.

After three, four very, very tough years, we are going back on track. Both the consumer sentiment and our finances are going back on track, which is really good, at least for me.

About our scale: five out of the 24 billion in revenue that we do yearly is through our digital channel, through our online systems. We operate digitally in more than 80 countries. And what is more surprising to me is that we engage every year with 1 billion different consumers in any of our channels. That's astonishing knowing that there's only 5 billion mobile owners in the world.

Where we are different is we have a unique hype. When we launch some of our most unique products, we move from operating at maybe 1% of the scale of Amazon to operate a thousand times bigger than Amazon for a couple of minutes.

All this required, over the last years, a very, very advanced technical mastery. We release like every three, four times per working hour with more than 99% success. And we measure, for the last three, four years already, our outages not by availability but by the business that we lose by every second that our system is out. So for us, MTTX, MTTR is just money.

And in 2021, we were doing this with 1,400 amazing engineers on the online shop, all our online different channels from marketing to social to online. We had — like in the picture — we had muscles everywhere.

But really, you heard, three, four really tough years as a company. Did we really need all these muscles everywhere, or did we just need, as a climber, to have muscles where it matters really, where it serves our purpose?

So we currently work with half of our capacity — we released almost all our external providers. So the Accentures, the Infosys of the world: sorry, you hate me, that's fine, that's fine. You still have a lot to do.

We try to work smarter, and we try to work smarter thanks to this community. But that's not all that looks like a challenge — it's not all on top. And thanks God, after pandemic, people are back on physical retail, and also our wholesalers that started their digital journey later than us. So JD, Intersport, Big5, Zalando are leveling up their digital game with 10% growth year over year. So we need to stay ahead of the curve by looking as premium as we look in our own channels, also anywhere where the consumers are. And also recognizing that the journey of the consumer is not in our ecosystem — it is everywhere. So when you want one of our products, you go from paid media, search, social, to our partner site, and you buy whatever you want.

And this is more important because what I like to say to the team is: our consumers — you don't know where our consumers will buy from us in a couple of months. You only need to see Samsung — the last refrigerator that they released in Japan is doing your groceries. So why would a person not buy adidas from their bike, from their glasses?

It was like a challenge, but as I said, thanks to this community, as we typically say, we got this. So I will go very, very quickly through five, six topics that we got from the community. And I'm linking the talks from smarter people than me that I typically bring to the talk if you want more details.

011. Find your business strategy

And why do I say finding? Because at any point you will have more strategic business leaders or more operational hardcore ones that just get things done. Again, my advice is: you need to look for something to hold on, something to anchor your technology strategy. Otherwise, your tech people is just working on the minus-one floor just to have fun with technology.

For us, now, in this moment, as I said, our business strategy is to create a connected technology ecosystem. And if you see in the middle, we anchor our composability journey into this business strategy.

022. To anchor your technology strategy

So we need our technology to scale both regionally and in special moments that I commented on. We are consumer-facing, so we need to be fast — stay fast. That's not a trade-off. I always put the example: we were the first one in China able to connect with TikTok in two months. 30% of our revenue was coming through TikTok. We did it in Europe — it was the biggest business failure ever. Two sales in a weekend. But we were there. We were there fast.

And yield — I don't want to talk about efficiency with our engineers. I want to talk about maximizing build value over running costs. That is just throwing money away.

And then we need our ecosystem to be smarter. So there is no way that we connect smartly with 1 billion different consumers manually. It's impossible. We need 1 billion people working. So it needs to be through data and AI.

And last but not least, as I said, connected ecosystem — not reinventing the wheel into the different channels, going to the basics, respect your foundations.

033. Build over solid foundations

The whole company is very complex. You see later, the whole company is working in product areas, in domains, including infra, data, platform teams. We can easily say that we were early adopters of what I hear here, and it became the standard of the industry. So our platform and our data teams working as products, while our consumer-facing product teams working as platforms for our business platforms, for our markets, to create experiences and reach our consumers in unique ways.

044. Maximize focus and value

Jon Smart — you, we had the pleasure of having you in our little town in Herzogenaurach. You observed how we are obsessed with maximizing focus on value by working in value streams. And for us, the value streams were the steps of the consumer journey, because it was the easier way of getting clear business KPIs of how we bring the consumers into the platform, turning them into members, how we enlight them with content, product, with what they want, how we make the purchase, how we deliver the products, and how we give them the best consumer service.

And then about visualizing and actualizing flow — the teams are, for more than six years, using Flow Framework for their job. But I think it's now, for the last year, when we are really actualizing it at scale at the executive level in two ways. So we use flow distribution for the capacity allocation. So cutting the team by half is forcing you to make choices, and you don't want to cut too much on some topics like data privacy, etc. And we even this year used the Flow Framework to decide which parts of your cost we are gonna capitalize — what is really run versus build in the box. So it was really a good exercise.

And on flow efficiency, after trying a lot of tools of connecting our different development tools, etc., and Jira, we went back to the basics. We just measure the time that the tasks are spending into each of the steps. And we are asking proactively from the management: what are the three things that you're gonna do every quarter based on the data? So it's actualizing over visualizing.

055. Visualize your architecture

And last but not least, we talk about enterprise architecture, but at the end, we're having more than 300 microservices in our ecosystem. So how do we visualize this plug-and-play microservices or capabilities for the different markets? We invented a common language with our business to visualize it — we call it the capability diamond. So we catalog every reusable component that we have, and you see on the left, into Packaged Business Capabilities, which is basically a set of microservices over some data layer that communicates with the world with some APIs and even streams. And then if you put on top the user interface for the consumer, and the admin interface for the user, is how we have the diamond.

I strongly recommend the talk from my head of architecture, Thomas Gilián, in Amsterdam — you can see in the QR code — and you will see how this is boosting reuse and speed at scale for us.

066. Always with your people

And the last point that I was gonna take is people. We are nothing without them. And for us, after years of embracing growth, communities, technical mastery, it's been the year of embracing change. Why? Because our product people, technologists — they love their products, their identities, their products, their technologies. And that's not good. That was good. But now we foresee that more than 50% are gonna need to change technology or products or focus in the last two years. So we need them to embrace change.

We track every quarter the health of the teams with regards to 10, 12 different metrics. '21 and '22 were for us the year of focus on going to consumer, and this led to our value stream transformation. '23, '24 are the years of learning and fun, because we want it to be fun.

07How did it go?

Thanks to this — and thanks to this community, really, I mean it — we can say very proudly that we are leaner. So with half of our engineers working in our ecosystem, we also decreased $25 million per year our running cost, and we opened in 20 new markets for the last two years. And more important, in our new architecture, every 10% growth in sales will only create 1% growth in running cost — one or 2%. So we are built for the future.

We are faster, definitely, in every metric. And we are smarter. We just released a month ago our new personalized home, and it's converting three to 5% better, better landing to PDP, how we call it. And our consumers are engaging four times more with it. And so we had to roll it out globally in just two weeks — no A/B testing, nothing — it's working like a charm.

08How GenAI is changing the world of adidas commerce

I'm talking about being smarter. So how is GenAI changing the world of adidas? It's changing it all around. You should only see the report from the WEF saying that 85 million jobs are gonna be replaced next year in the world. I prefer to say that they are gonna be improved, because I start serving that. In adidas it is changing everything: how we create products, how we get inspired, how we run our consumer service and talk to our consumers, how we sell our products, how we generate stories, how we create content.

But, of course, as I said, we have a champion mentality in technology, and we had to test it, we had to try it, we had to use it, we had to embrace it.

So how did it go for us? This is an exercise that we did Q4, Q1 last year. And I will not call it a pilot because it's a fact. In the first wave, we put it to motion — we put Copilot to 500 engineers. And the satisfaction has been astonishing. Not only to search for information but also to complete repetitive or not-repetitive tasks. But what about the quantitative data? 82% — so it's 400 people — of the 500, they are using it every day already. So please don't remove this from me — we will never do. 91% find it useful.

And two-thirds of them really increased their quantitative metrics. So they do more commits, they review more pull requests, and every pull request 19 minutes faster.

But now is the question from our management. Yeah, but how much time are we saving? How much money are we saving? You know, our management is not very, very concerned about technology. Our engineers reported an overall improvement of 15 to 20%, which is very aligned with what we heard from the industry, of the time coding and testing, the time on IDE. Yeah, but how much time is that?

We took a look into Gartner, and I was scared seeing that Gartner is reporting that only 25% of the time of our engineers is doing engineering. I could not believe that. So we are hiring people to do engineering and they are spending only one-fourth of their time doing that?

So we ran an analysis, and we brought seven teams to track their time for one month. It was seven teams across four different domains, complete different maturity and technology stacks. We wanted to focus on how much time they spend on pure IDE — is coding, testing — is this really 25% of the time only? Then how much time did they spend into other product generation tasks, other useful productive work like documentation, analysis, design, etc. — we call it time on keyboard. And then how much time they spend in huddles, taking coffees, trainings, team ceremonies, etc.

And it's an exercise that we did already back in 2018 with a couple of teams when we were much more unmature.

So two takeaways. Our coding time is 11 points higher than benchmark. I still think it's too low. Honestly, I'm not quite happy with the results. And our time on keyboard, which is what I'm more proud of, is 18 points better than in 2018, when we were much more unmature. This is what makes me more proud.

But last but not least, when you look into details, there are clearly two different groups: the high performers, the high-productive teams, with more than 80% useful time, and the rest, with less than half of the time. What is the correlation? The correlation is very clear: with the seniority of the team, with how mature is the DevOps journey, and with how loosely coupled and open technologies they use.

But at the end, if you look into the four teams that are not high-performing, honestly, they are not high-performing because we don't want them to be high-performing. I know the teams, I know where they work. It's good how they are, and that's all.

So my two things: our level of transformation paid off. So thanks to this community, and paid off both in good times and bad times. GenAI tools is already a commodity for the team. So you cannot remove this from them. We will not see any tutorial, any blog post on technology that is not using it.

And last, embracing change, working on your ways of working, is even more impactful than GenAI.

My help is really asking: how are your consumers, internal and external, embracing conversational user interfaces — mainly human-to-machine? Are they shifting from transactional applications into conversations? And if any of you believe — I was having a nice conversation with Jason into also machine-to-machine conversational interaction over transactional APIs — yes, come to me.

Thank you. Thank you.