Host Intro (Gene Kim)

So would you believe that five years ago, the team from adidas attended this conference back when it was in London, and they left inspired after seeing a presentation from Jason Cox. This actually led to the creation at adidas of a centralized technology group to better advance the goals of every one of their business units through engineering excellence.

Fernando Cornago, who is now VP of Digital, has presented at this conference for the last four years describing the progression of their incredible journey. Over the years, Fernando has presented on his ever-increasing responsibilities as they kept creating value that was appreciated, how they responded to the global pandemic, which included both cost control and a massive focus on improving their e-commerce capabilities when it became the most important revenue channel, and so much more.

Each year he talked about their new challenges and also some of their setbacks. So here to describe the most recent parts of their remarkable journey is Fernando, who will be presenting with Thomas Gieling, Senior Director of Domain Architecture. So here's Fernando and Thomas.

[Music plays.]

Fernando Cornago

Well, hello everyone, and welcome to Amsterdam, the home of adidas -- of adidas e-commerce. And you heard: every little thing is going to be all right. And I know that we are all going through some troubles here and there.

My name is Fernando Cornago. I have, as I always said, the best job in the world. I have the pleasure to lead, for the best sports company in the world, all the technology of our consumer touchpoints.

And yes, I was indeed inspired by Jason five years ago, and I try to come every year to give back to this community, but I always get more than I give. And this is why I will continue coming back.

So I'm here today for a very complex topic: how to compose and visualize the technology capabilities that are going to make us win in the most complex time in adidas in the last 30 years, said by people here longer than me. And I am lazy. So I tried to ask the GPT version of three of our influencers -- Messi, Pharrell, or even our founder -- but I didn't like the responses, so I brought the smartest person in the team. And who is the smartest person in your teams? Yes, your head of architecture. In this case, it's not only a cliche, it's also true.

So I bring with me Thomas Gieling. He is the head of architecture of Digital for adidas, and he was here when the first line of code of e-commerce was put in the system.

A little bit of context first. Within the last seven years, our push for digital and direct-to-consumer made us play in the top five, ten mono-brand e-commerce in the world, with five billion digital revenue and 360 million members across the globe. One third of the WhatsApp users of the world are members of adidas. And we are there with the likes of Apple, Zara, H&M, etc.

But I always like to say that we are more fun than them in three aspects. One is our variability of load and throughput. We typically operate a hundred times lower than Amazon -- it's Amazon. But during the first couple of minutes of a hype drop, when we launch our top collaborator with Stella McCartney, Gucci, Balenciaga, we have a thousand times more load than Amazon during a couple of minutes. Imagine the pressure that this puts on our systems.

We operate also in 63 markets, 363 countries, with different consumer expectations, delivery methods, payments, privacy regulations. And last but not least, we are more fun than the others because we have much more products than them, which makes up our store sometimes look as a bazaar. But also we are targeting almost every sport, 20 key categories, and every single persona: super-value buyers, family shoppers, super-premium buyers, etc. All this requires technical mastery.

Over the last year, we've been doubling the number of releases to production over the years. Like yesterday from HSBC, we deploy to production 4,000 times a year with a 90% success. We are very proud of it. Every hour that our system is down, we are losing one million euros on average. And we know this is much worse in big sales, Christmas, special sales. We know that our competitor from Oregon last year lost in four hours more revenue than we in the whole year because of a complete outage in Amazon East. Luckily, we are on the west side.

And we've done this with 1,400 engineers at peak. I will tell you later why this is at peak.

Over the last four years, I've been telling you around the non-negotiables of our engineering strategy and tech strategy. We have a fully integrated tech hub location strategy, where every one of our seven tech hubs where we build software from has a clear global goal and responsibility. We are not throwing people there just for the sake of throwing people in any of these places.

You don't differentiate who is an engineer, who is a product owner, who is an analyst, or who is a business person. They serve the same targets that we cascade from the company five-year strategy into our OKRs. A lot of education for our management: OKRs are ambitions. They are not targets. I don't want to see everything green. That's not good.

And last but not least, last year we moved into our value stream delivery strategy, or delivery method, where we align our teams and our targets across the consumer steps of the journey and also around the business users that are using that.

This made our engineers, for the first time, increase their perception on speed. We got two goals with that. They value themselves nine out of ten in purpose and value, which is crazy coming from an engineering team that I always said was coming from working in the minus-one floor, super far away from the business. And second, they moved after three years from seven to eight in speed, which I don't know if it's good or bad, because I also don't want them to consider that they are ten out of ten in speed. I always want them to want more.

And we heard Maya yesterday, an amazing presentation. We all started by culture. If you work with us, with adidas, you will see a very marked character or culture with the mix of sports and engineering. It is clear in the first conversation with anyone from our team.

So what happened in 2020 is that our world completely changed and the rollercoaster started. Pandemic: we had to put all our efforts into e-commerce. It was our only open store, but we didn't have inventory to sell. Then first-party restrictions and different privacy restrictions in Korea, China, Europe, U.S. When everything was going better, then China locked down, the war in Ukraine and Russia, inflation, we now had too much inventory, and there were more discounts than ever. It was crazy.

And then our goal of technology was always: be fast and stay fast. The example: we were the first brand able to integrate with TikTok in China for livestreaming commerce in two weeks. We were integrating our products, our descriptions, our data. In a couple of months we were doing almost a third of our digital business in China in TikTok. That is one tenth of our digital revenue in the world in a couple of months.

You never know what a new consumer is going to want, and you never know what is going to be in the next channel or the next platform.

This is how we created this mental model. I always tell my team members -- and I see three or four here, because they were just around the corner -- that they should put this below the pillow when they are in the bed. In order to create value, we should not create any piece of software that is not based on data. Our search needs to be fed by searches that are happening. Our recommendations should not recommend anything that is not happening, inventory that is not purchased afterwards, etc.

After that, we don't create value if we cannot go faster to the rest of the markets. It doesn't matter if something is working in the UK if it cannot be deployed to the rest of the world tomorrow. We will be late.

And last but not least is adoption. Nothing that we create creates the full value until it's adopted across the market. And the best way of things being adopted is that they are evergreen, that you don't need anyone operating them.

But things got worse. They got even worse. The week after I was in Vegas in October, the world blew up even more for us. We lost our most profitable collaboration. It was 5% of our sales, but 30, 40% of our profit. Retail is decelerating, and commerce is still higher growth.

So we found ourselves with a higher operating overhead, one that we could not afford. And we have a new CEO, who I think is doing pretty well, but he's a marketplace guy. He wants us to be everywhere where our consumers are, in every channel.

Our challenge was: how can we keep mastering technology in our complexity with half of the team? We released in six months all our external parties. And the good news is we keep providing the same technology performance so far. Our number of releases stays stable. I don't think we will double this year. Same success, and our stability and security KPIs are there.

The first thing to change is the mindset. You should not have muscle everywhere. You should have muscle that helps you to achieve your goal. Who is scaling better? I think the guy on the right.

So to our framework that is making us fast and is silently making us better. Any of the 4,000 releases now are much more impactful than the 4,000 releases of last year, because we have better code, we have less technical debt, everything impacts everywhere.

We added a couple of accelerators, like our capability strategy and our capacity management and monitoring, as accelerators for the transformation and shrinking our teams.

Capability strategy: we have people that are in love with their code, in love with their product, in love with their purpose. But maybe what you are doing in your product is diminishing returns. It doesn't mean that your five years were thrown to the garbage. It was fantastic, but we don't need to invest here anymore; we need to invest there. It depends on our focus on the capability and also on the maturity of the capability in the market.

And second -- and Mik Kersten was here yesterday, I don't see him today -- using the most basic of the flow metrics, flow distribution, into monitoring where the teams are investing their time. There is no benchmark that is going to help you better than observation. If one team is divested, you need to see: is the team able to provide value or not yet? Is it linked to ours, or should we invest somewhere else? Is this team struggling even to be compliant because they cannot keep up with the risk of this market?

So at the end, before handing over to Thomas to tell you how we are doing this, it is all about using and leveraging technology first. I always use an example from Fosbury. He was, in 1968, a high jumper, but there was a new technology. They put, for the first time, foam in the landing zone. And he was the first one that instead of jumping scissors, he jumped on the back. He was the first one using technology to work differently.

And now we are on the edge of changing with what is coming with generative AI, DevOps, etc. The ones adapting it first will be the ones winning. And in order to educate our business with all these changes -- because imagine, it's been a tough six-month journey -- Thomas is going to bring a tool that we are using for visualizing technology.

Thomas Gieling

Thank you, Fernando. In order to explain what we do, first I want to take a step back and explain what capabilities really are and how we define them in adidas. So when we speak about capabilities, we speak about packaged business capabilities, which are software components that represent and perform a well-defined business capability. They're centered around data models, events, and APIs, and they act as a single entity.

We start with a data model, and we define a data model. We expose that and then we implement that data model in several microservices. Those microservices expose APIs and events, and they have an optional user interface.

In order to start leveraging these packaged capabilities on their own, they don't implement a complete user experience. So to look at what is the consumer experience that we want to bring to the consumer, but also how do we administer that once we've brought it live so that once an experience is successful, we can actually scale it, we start looking into what are the capabilities involved inside that experience and how do we compose them together in order to generate the desired experience.

This allows us, at a very early stage, because we're still talking business architecture at this stage, without doing an architectural deep dive on how do you implement this, what are the changes required, to understand what the complexity is of the changes that we're trying to make. And complexity drives the time that it will take us to deliver that.

As an example, we have our payment capability, where we have several microservices that abstract the payment vendors that we use, the payment providers. They all expose a standardized data model of what we at adidas consider a payment is. And that, combined with an orchestration, exposes a single set of payment APIs and a single set of payment events. That does not provide us with a payment experience, because this is just the back-end service that provides that.

To provide a full payment experience, you also need fraud service. You need to persist the payment information inside a basket, and you also need to create an order once that payment is completed. If you look at the next step, then you see that together with this payment service and composing that with the other packaged capabilities, you can deliver this final experience to the consumer.

Similar to our inventory service, we have a lot of various inventory spaces because we operate stores around the world. We have partners that have inventory that we hold. We have third-party warehouses, we have our own DCs, we have smart replenishment systems. All of these capabilities provide our inventory service for the inventory team. This is their full diamond.

When we look at our product wayfinding experience -- allowing consumers to find the right products at the right time in their journey -- we also rely on product data, we rely on intelligent search, we rely on digital asset management and personalization, and within that we also rely on our inventory. So what is a diamond to one team can also be a packaged capability to the rest of the organization.

What is more, this diamond for product wayfinding, because it is a composable setup, can be used with other heads as well. Seamlessly, you can use those same APIs, the same tools and technologies that you deliver for your product wayfinding experience in the digital ecosystem -- so digital direct-to-consumer -- but you can also leverage that in your partner commerce as well, so that our partners such as ASOS, Zalando, JD can leverage the same capabilities and the same technologies that we use for shipping and getting the right product information to the consumer inside the partner programs as well.

So what do these capability diamonds and the capability-led architecture give us? They give us structure in a microservice architecture. Very importantly, we have hundreds of microservices, but we only have around 20, 25 capabilities. Identifying which team do I go to to get my changes done, identifying which roadmaps do I influence, becomes a lot simpler.

It also allows us to have discussions on complexity at a very early stage of the design process. We don't need to go into an architectural deep dive. We don't need to get involved with 15 people from 15 different teams to do something simple. You can just quickly ramp up a small group of experts and understand what the impact will be.

It also -- and that's probably one of our biggest benefits -- creates a common ground between business and tech because the packaged capability is defined on a common tech and business understanding of the process. You can start having design discussions with your business teams without it getting too technical for them.

And also it helps us avoid duplication and encourages reuse, because it's easier to find the data that you need access to because there's a lot less data or capabilities that you need to search through. It's easier to find the right tools that you can use, and it's easier to reuse them as well.

Back to Fernando.

Fernando Cornago

Thank you. Thanks, Thomas. Brilliant. I love really coming here to the headquarters and I start talking with my business colleagues. It's fantastic.

Yes, to close: Gene asked me, what are you doing with AI? We heard a lot about generative AI in engineering, architecture, etc. I'm going to tell you how generative AI is changing a little bit our world.

Zalando, by the way, just announced that in a month in Europe, they're going to pilot a generative-AI-based shopping system. Let's see. They're going to get the first back shot. We will be closer to them.

We are already using generative AI to inspire our product designers. We have our private instance where we are mixing our current product DNA, trends in the markets, etc., and inspiring our designers.

But last but not least, where I think that it's going to completely change our world is in content generation, personalization and storytelling. Not only because it's going to get better, but because there is no way to scale in a personalized world where you need to tell the stories differently to a family mother, a father, to a Latino-origin person like me, to a super-trendy Gen Z. They expect different things.

Imagine your kid, when you are purchasing a kit for your kid, that he can see himself celebrating with Messi the last World Cup. Or imagine if you are running on the desert, that suddenly when you look in your product, you see yourself in a story where you go to a forest and you are rescuing a little kid that is in trouble thanks to your running jacket. This is definitely the future of generative AI for us.

Where do we need help? If any of you went through such a sudden cut in cost, efficiency, etc., and you survived it, please let me know after that. It has been tough, but the team is fantastic and we keep performing at the best.

Thank you.