Host Intro (Gene Kim)

[00:00:02.180] Fernando Cornago is the Senior Director of Platform Engineering at Adidas. I had the pleasure of meeting him three years ago when he presented with Markus Rautert, his VP, about the group of rebels who were bravely seeking to create a better way of working inside of Adidas. They had regular meetings at this place called Cafe Intencion, where they made and executed their plans, eventually leading to a meeting where they pitched the CIO at 10:00 PM at a leadership offsite, which led to the creation of the platform team.

[00:00:27.640] He will be presenting with Ben Grimm, Senior Director of .COM, who drives the product vision for the e-commerce channel that generates billions of euros of revenue annually. They will be presenting on the continuation of the Adidas journey, including being responsible for one of the 23 board-level initiatives, and how they're educating people across business and technology on creating world-class practices to help them win.

[00:00:49.600] And by the way, I was so inspired by their story. Whenever in The Unicorn Project you hear about the Dockside Bar, that is indeed modeled after Cafe Intencion. Fernando and Ben.

Adidas Brand Video

[00:01:06.500] If we can see things for what they are. If we can see what they could be. If the plastic we use, we never throw away. If the end of one thing could be the beginning of the next. If we know that less can create more. If we can return, we reciprocate, we regenerate. If we are here for others. If we can work as a team. All we have to do is connect, and the world opens up. The future is about giving back.

Benjamin Grimm

[00:02:10.200] Hello, everyone. Thank you for having us. I'm Ben, and with me today is Fernando. We are both working at Adidas. Thank you, Gene, for having us today in Las Vegas. It's so exciting to be here, and also thank you to the whole team for the organization.

[00:02:27.220] As you have seen in the video, we at Adidas, our core belief is that through sports we have the power to change lives. To do that, the best way is to build one-on-one relationships with our consumers, and digital helps us with that through premium, personalized, and connected experiences across all our touchpoints.

[00:02:47.700] In the last years, we have seen premium experiences and new enhanced experiences coming up across the whole industry, especially around the sports industry. Our target groups have high expectations toward those experiences across touchpoints: very individual, very personalized.

[00:03:10.440] We just opened our flagship store in London, which is the best retail digital experience we have ever built so far. When you are at the next DevOps conference in London, you can definitely come by and have a look. It's really amazing.

[00:03:23.860] About two or three years ago, we had a good basic online store, no app outside. But today it's a different marketplace, and the opportunity and the potential are endless beyond 2020. Today we talk about our platform strategy and our latest progress. Fernando can give you all the insights, and I will talk more about how it enables our business and our growth, and how we achieve the targets with that. Overall, this is how we transform that 22 billion euro net-sales company toward a future-ready digital company.

Fernando Cornago

[00:04:06.260] Thanks, Ben. Hello, everyone. Good morning. It's a pleasure to be here in this crazy city for the first time in my life. My name is Fernando Cornago, and I'm heading up the platform engineering practice for Adidas. As Team Topologies would define us, we are at the same time a platform team providing self-service platforms to our engineers, and an enablement team helping them learn good practices and the usage of our platforms, with the final goal of making them autonomous.

[00:04:32.969] My relationship, as Gene said, with this fantastic community started three years ago. I was the lead of global software development. We had a horizontal development model for Adidas, and I came to London looking for inspiration, and I found it. I need to thank especially Jason Cox from Disney, because he made me see the light. I came back directly to our small German village, Herzogenaurach, where our founder created the company 70 years ago. He was born much earlier.

[00:05:08.660] I started a couple of meetings with different technical leaders in different departments. We were meeting in a small bar we call Cafe Intencion. Our goal was to convince our CIO and our business leaders that software and new practices should really be taken in Adidas in order to make us successful. We were successful, and I was talking about that last year.

[00:05:33.360] This year, Gene asked me to talk about culture, strategy, and how to bring our business together. That's why Ben is here with me this year, because he's the owner of our biggest, fastest, and most profitable store in the world: our .COM.

[00:05:51.300] We go every day to the office really wanting to change lives through sports. But basically, software is changing sports. Not only how we practice; nowadays no one is going running without a smartwatch and then sharing the activities with friends to gamify the experience. Also how we consume it. We are in the U.S., so watching an NFL game right now is a full experience of real-time stats and probabilities. It's amazing.

[00:06:16.944] Are we a software company? Definitely we are not. All our revenue comes from physical products from the different channels. But we need to start behaving like one, because we have more than 250 million lines of code in our code repositories that are accessed every day by more than 2,000 people.

[00:06:37.684] In a new world where we push product ownership to our business because we are organizing value streams, what is the role of technology? For me, it's having the best engineering practices to provide speed at the right quality: sustainable speed in the long term, but really speed to succeed and fail and change our way.

[00:06:58.924] Especially us, because we saw from the last State of DevOps from Nicole this year that retail, where we are in, is the only industry that correlates directly with top performance in IT. So we are there, or we are dead.

[00:07:17.004] How do we do that? We go DevOps by default. That is the strategy. No one is challenging that. Not all the teams are there, but we go DevOps by default. We define DevOps, with inspiration coming from my first year here, as Jez Humble said, with the word CALMS: culture, automation, lean, measurement, sharing. This is what we use for every single one of our activities.

[00:07:46.744] Then it comes to platforms. Can we do DevOps with a monolithic architecture, with a commercial off-the-shelf system? Definitely you can do some steps, but there are certain capabilities of the platforms that need to be there in order to boost and help teams go easier into the DevOps journey.

[00:08:05.344] The platform has to be on-demand and self-service. Ideally you want to run it up and down on automation, command line interface, SDK, API, you name it. It needs to be scalable and elastic, so it can scale up and down depending on the needs, ideally based on observability in an automated way. You need to pay per use. It is not the platform team who is having the budget. We don't want that. We've seen a lot of times when infrastructure has grown like hell and no one knows what is there because the teams using it are not paying for it anymore. The teams should pay for what they use of the platform.

[00:08:43.884] Transparent and measurable observability is key in two directions. The application team needs to know what is happening in the platform in real time, and for the platform team it's good to know what is happening in the team, because sometimes we think the platform is up, but the applications are not. Last but not least: open and inner source, not only using open source by default, but opening the platform to the community so they feel part of it. You open your backlog, and even if you don't have capacity, you open your code for others so they are contributing to the platforms.

[00:09:18.724] How do all these technical things fit into our strategy? We have a pretty focused CEO globally, Kasper Rorsted, and he has ranked our more relevant projects up to 23. I have the luxury of running one of them: Data Tech Foundations. Data Tech Foundations is helping all our core data initiatives, from digital commerce, how we sell, and Ben's team is dealing with that; how we engage our consumers and membership programs; how we deliver in an omnichannel strategy; and how we create end-to-end data creation. That is like the meta lead time. By shortening the lead time of software, we enable the company and our business to create physical products faster because we create digital first. That helps us in our relationship with factories, resellers, wholesalers, and B2B business. They don't need to see the product anymore in order to sell it.

[00:10:26.804] The fact that this initiative is relevant is super cool. We are Adidas; we make our revenue from physical products. We need to present this to our board every three months, and it's helping us in two directions. We are educating the board in things like automated testing, making data easily consumable, continuous delivery, number of builds, and quality. At the same time, this forces us to link everything that we do in the bottom of engineering, in the core engine room of the company, with business results, with direct money if possible.

[00:11:05.404] Sometimes we struggle in explaining why quality is important and why different capabilities are important. But then we realize that what we are doing in the platform is not that different than what Adidas started doing 70 years ago and what has helped our company be successful all this time in a leading position in the sporting goods industry.

[00:11:31.624] We call all this the Adidas Digital Platform. It's nothing else than a set of capabilities that work together and allow our teams to go faster. Let's start with ODP, Open Digital Platform. That is the compute part. It is where our interactions with consumers happen.

[00:11:49.384] If you go to one of our stores, seeing a big queue is super frustrating. Wouldn't it be cool if the store assistants popped up when the queue grows and popped down when the queue disappears, even going up to zero? We can do this thanks to our more than 29 Kubernetes clusters and now with serverless capabilities. Wouldn't it be cool also if these store assistants are popping up wherever in the world we are? We need to be closer to our consumer. Nowadays, for Adidas, the main focus is North America and Asia. I come this weekend directly from China. If you want to be at the forefront of digital retail or Industry 4.0, you go there. The direction has changed. It is not global pushing things to China; it is global learning from China. For this, we have Kubernetes clusters. Our platform is spinning up in five different regions of AWS.

[00:12:48.708] Last but not least, wouldn't it be cool if these store assistants are popping up at home when you cannot leave? This is why your platform needs to run on-prem for applications that cannot leave on-prem because of data confidentiality or mainly data gravity.

[00:13:02.528] Data is where all the innovations that are coming in the present and future will come from: getting access to the data and playing around with it. We generate gazillions of data from product design, manufacturing, supply chain, selling, delivering to a consumer. Accessing this data quickly is key.

[00:13:24.968] We see the data as the baton in a relay race. The runners are small DevOps teams, and they need to be fast. But the four fastest runners do not always win the race, because the handover needs to be smooth. We achieve this thanks to our global streaming and API strategy across the globe, where, thanks to smart contracts between the different teams, you can access the data from another team quickly, and it is cataloged in our data portal. At the end, where is all this data flowing? It is flowing automatically to our big data platform, where we leverage all the power from our hyperscaler, AWS, so our data science can play around with AI, ML, and deep learning where our normal algorithms cannot get.

[00:14:17.648] A lot of technical pieces. We said we are working on this to bring value to our business. Nowadays, building an app is putting together a set of pieces of a puzzle, and they need to work together at a certain point in time. Exactly like American football. The team has a purpose, a vision, and a goal, which is the touchdown. But for this to happen, the different players need to move at a certain point in time.

[00:14:45.648] For this, we use our implementation of Jenkins on steroids. We use Kubernetes to spin up ephemeral master agents for the different teams, so they are fully independent, install plugins, and run on their own. The platform team only provides backup connectivity to our systems and security.

[00:15:04.128] Deploying continuously is nice, but you want to deliver something that is working and analyze the user. This testing obsession we get from Adi. Our founder was a pioneer in going to the field and testing with athletes our products. We do the same. We need to test where the consumer is, so our automation testing platform is also spun up across the globe. Our physical product owners test products with automatic kickers and so on. Same concept with test automation for robotic process tasks in the back office and global business centers, tasks that are not making us different from our competition. We use UiPath also at scale.

[00:15:51.908] Last but not least: observability. No one wants to see a sport without a scoreboard. When I go to see my kids and I don't know who is winning, it's not the same. You need to know what is happening. We use the same approach with data that we use for metrics. All our metrics, traces, and logs are cataloged and put in an operational lake so they are accessed by other teams. Whoever wants can access metrics from a dependent application.

[00:16:26.528] Technically all this is cool, but nothing will work without people. People are our biggest asset, especially in engineering. A platform is as successful as it is appealing for the teams. They need to feel that they are part of it. This is why you need to boost community, training, and a lot of activities so they feel part of the game.

[00:16:49.508] If there is something I feel proud of after these two years of journey, it is the community we have created. Adidas code: you can see it on our T-shirts. We started with no engineers four years ago, and now we have more than 300. We are now doubling this number by opening up new centers in Colombia and in India. We have more than 100 or 150 events a year, with even one day where every single engineer of Adidas is in our headquarters for an internal conference with internal and external speakers, hackathons, everything.

[00:17:26.588] We have a smooth onboarding. Every single technical person joining the company comes to Spain, where currently 70% of our engineers are, and they build an application end to end. In two days, in the last two days, they build an application, when in the past it took us even six months to get it to production. Most important is that they meet all the different responsibles for the platform, so this community is built forever. That is the most important thing on the onboarding.

[00:17:53.538] Then culture. I'm not here to talk about suggestions, but what I do and what I apply is applying your company culture in everything that you do. I come from different companies, and I would never have done this in different companies. But in Adidas, we did that.

[00:18:11.708] In the monthly meeting of all the directors of engineering, we were struggling because we had everything: the DevOps maturity framework, all the accelerators, everything. We saw some teams were not progressing as we wanted. But we hate top-down mandate in engineering, and even more in Adidas. In Adidas, you put a rule and someone is jumping. So basically we said, "In Adidas we like competition. Let's open a cup." The only thing the team needed to do was say where they were and where they wanted to be in nine months. The only condition was to bring their product owner and tell us also how this brought business success to the team.

[00:19:00.488] Participation was amazing: 220 people applied, and every team received a mentor. The coolest stuff is that they received a mentor from a different team, so both the mentor and the team received fresh perspectives from a different vision of the world. The winner, the mobile team, is doing a world tour around all our different headquarters: China, U.S., Russia, to really spread the word.

[00:19:29.468] The adoption of the platform and best practices grew like hell. We have more than 10,000 Kubernetes pods running now as I speak, and more than 116 projects in our software delivery lifecycle metric tool that takes metrics from all the different tools: Jira, Sonar, Bitbucket, Jenkins, and then creates metrics for the team to improve. Even all the critical teams are using Chaos, our chaos environment, and DevSecOps in their lifecycle.

[00:20:00.608] Golden metrics: deployment. Ben's teams went from six weeks to deploy to basically whenever they want. The mobile team, on cycle time of a campaign, went from three days with lengthy emails and long meetings to a button for the product owner. The consumer service team went from 26 outages a quarter to zero, all thanks to improving their DevOps practices.

[00:20:27.828] Business value was amazing. The Adidas app team started with six or seven people. They reported about 3,000 minutes freed for working in real features. They scaled the teams up to eight teams, and they grew the revenue by five times during the DevOps Cup in nine months. Consumer service, running on Salesforce, pushed Salesforce and challenged Salesforce on integrability of their code, testability of their code, et cetera. They turned themselves from being a cost center into a value driver, turning complaints into business opportunities and conversion for our e-com.

[00:21:15.128] Product creation is a perfect example of platforms over platforms. We provided them with the platform for compute, data transfer, and APIs, and they built an operational data lake for product creation. They are an enablement team, so they have a goal to disappear. Now they are disappearing, and they are pushing all this information and ownership to the teams. At the same time, last year they already saved half a million for this company, a team of six or seven people. Store associate mobile, providing mobile applications to our store assistants, grew adoption multiplied by 10 because of trust, because of being able to deliver fast and in a secure way. They even reported that they saved a good bunch of trees, which is always good for sustainability if you save paper.

Benjamin Grimm

[00:22:08.768] Thank you, Fernando. I'll take the next couple of minutes and talk about how it accelerates our business, because as Fernando already said, at the end it's all about speed. We see this as a competitive advantage: to invest into the platform to get that speed.

[00:22:27.008] Looking back around three and a half years ago, Adidas came together and created a new company strategy called Creating the New, where one initiative out of that is the platform. In general, there was a very strong agreement in investing into digital. Where we are at the moment: in 2018, from that 22 billion we had 2 billion net-sales revenue from digital channels, and we would like to double that by 2020. The expectations are high toward technology and the growth.

[00:23:03.408] Back in 2016 and 2017, which sounds like a long time ago but is just two or three years, we started with a very basic shop. We had no app, we did not work in social, we did not do good email communication, and so on. Now we have more than 100 engineers working on the experiences across our portfolio and ecosystem.

[00:23:27.388] When we set our objectives and key results for the quarter, we want to stay flexible in how we achieve our outcomes. The platform and the teams give us the flexibility to populate teams across the most important initiatives, and they can change quarter by quarter. It can be a new app, or a new experience to sell exclusive products to our target group.

[00:23:58.408] It's faster time to market as teams don't need to focus on building basic capabilities like monitoring and reporting. Scaling and performance issues were a problem in the past, so we don't really see that anymore based on the platform. It's easier to scale. We have around 1 billion views a year on .COM, and we see the improvement in cycle times and lead times around the development cycle overall.

[00:24:32.472] What is important for us is quality as well, especially when we go into the holiday season now, with 11.11, Singles Day in China, which is bigger than Black Friday and Cyber Monday together. It's a huge day for us, plus the end of November, Black Friday, Cyber Monday. Thirty-five percent of our business is in Q4. Usually we don't think about that: sports, summer, spring. But we do all the main business, especially in digital, in Q4. Imagine you do 4 billion out of 22 billion net sales. It is getting very significant in the last two to three years.

[00:25:13.572] At the beginning, if you have hiccups and lose some money because you have incidents, that's kind of acceptable and you need to work through it. But with these numbers now, we need to have a very strong and solid platform in place. There is no choice. We need to invest into it, and I'm so happy that it's a board priority.

[00:25:36.712] DevOps helped us with site speed. Based on the platform that Fernando just talked about, we created a new headless e-commerce architecture based on APIs with a new front end. We increased our load times from around four or five seconds to below two seconds, which is a very strong driver in conversion rates. We did that last year. Our conversion rate overall in the shop raised by about 10%, which was way better than every other feature just by improving site speed. No new features, no new products, just site speed, which was very successful.

[00:26:18.712] Last but not least, we want to create premium experiences for our consumers. Here are some examples we did in the last weeks and months. We see now the return on investment of the platform and the headless architecture. It takes some time to get there, which makes it very important to be consistent and transparent in communication toward the board. Once you get the return, it's very nice.

[00:26:46.432] The first example is a feature we just launched on the product listing page where you can buy inside the product listing page directly. You can add products to the cart. The imagery is head to toe, and you can click on an image and basically buy every product that is on that picture. The second example is about how we release products and tell stories: the photography is larger, it is all about the stories, and consumers can set reminders to be informed when the product finally drops.

[00:27:24.272] Another core commerce feature is checkout. It sounds pretty boring, but it's a huge conversion driver if you do it the right way, and it is a very complex beast. Around the globe, with different carriers, payment methods, and risk management, it gets very hard. We just released our new checkout in the U.S. one and a half months ago, one month ago in the U.K., and this week Germany. We see really significant improvement in the cart-to-order conversion rate.

[00:27:57.712] The fourth example is telling stories on product listing pages. For us, it is very important to do it there because there is the consumer. The consumer is not on campaign landing pages and other pages like that. They are on product listing pages and product detail pages. For example, for the product Stan Smith, we tell the stories directly on the product listing pages, which is very important for us to do.

[00:28:23.212] What I just showed you is maybe good now, but it needs to be good in the future, and we need to continuously invest into the platform and what is next. Fernando can talk about what is next.

Fernando Cornago

[00:28:40.952] Thanks, Ben. I'm going to rush a little bit because it will not be the first time that I'm kicked off the stage by Gene. What is next? DevOps is never done. But the teams are already too progressive or too in the game to really create another DevOps Cup. It is a proposal coming from one team: we are creating what we call the DevOps Maturity Incremental Index.

[00:29:10.792] We have a checklist of 70 things grouped in development, observability, product architecture, culture, and process, with different maturity steps from crawl to run. Now what we ask the teams is to report monthly on which or how many points they have improved, and how this is bringing business value. We as a management team only need to review every quarter and move some things from run to walk, and some things from walk to crawl. What is good today may not be as good tomorrow.

[00:29:39.232] Aspiring engineers: our role in a future of IT in an engineering-based organization is to have the best engineers. We created a progressive framework for learning. Each team catalogs the technical skills they need in different levels and the trainings to get there. It's good for Adidas engineers because they know what we expect from them, they know where skills are more hyped or more needed in the company, and they have people to help them get there through mentorship or even job rotation. It's good for the company because we can declare future skills: AI, ML, things that maybe we don't have very extended today but will have in the future.

[00:30:22.382] Gamification: we are a little tired of black T-shirts. We created a complete new internal platform for gamification where you can get a sticker if you are good in different technical capabilities: streaming, security, Kubernetes, serverless. You get one, two, or three stars depending if you are good, very good, or if you are even teaching others and contributing. You get these stickers, and if you get enough stickers, you can get more colors, a white T-shirt, or even hoodies. This is also working like a charm in Adidas.

[00:31:02.612] Last but not least, what is the next big thing? We want to copy the DevOps Cup and create next year the AI Cup. We are upskilling all our engineers in the next six months in AI and ML so next year we can launch, thanks to our product owners, an AI Cup company-wide. I hope to be here next year telling it to you. That's it. Perfectly on time, Gene. You cannot say anything. I hope to see you next week. Thank you.