 Good afternoon. How's everybody doing? So this is my first time at Slush. I thought I'd bring some sparkle with me. My name is Stacia Carr, and I am director of engineering for the size and fit team at Zalando. Today, I'm going to share some stories with you about how we tackle a really big problem with a fully cross-functional business unit. Now, you might be asking yourselves, what is a fully cross-functional business unit? How many of you in the audience are actually founders or working in a startup? Can I get a quick show of hands? So most of you. So what we're describing here is actually what looks like a startup. My team is 24 people. And what we mean by fully cross-functional or interdisciplinary is that we all have different skill sets. We've got data scientists, research engineers, product folks, designers, business people. And what unifies all of us is that we're tackling the biggest customer-facing problem in the fashion world, which is finding the right size and fit. About one in three of all of Zalando's purchases or return purchases are returned for size-related reasons. And whether you're shopping online or offline, you know it's hard to find what fits. Today, we offer our customers two algorithmic products that are size advice. One is a personalized size recommendation for customers that shop with us regularly. We can tell them exactly what size to buy. And the second product is a more generalized piece of size advice that tells customers if an item runs a little bit big or small. So over the last couple of years, my team has put a nice dent in this challenging problem for our customers. And what I'm going to share with you is how we did it. Before we came together as a team, we actually worked in separate parts of the Zalando business. We had business people working in one office in East Berlin and the tech people working in another office in East Berlin. Business people were typically running analysis to understand why items were being returned. They were really focused on numbers, very sales oriented, and also looking at our supply chain. So how do returns actually impact our logistics? The tech folks, very data science driven, trying to understand what can we do with the data that we had. And when the two teams actually tried to work together, it was really a struggle. They didn't have a single purpose that they shared. It was a lot like a tug of war. The other challenge that we had was that both teams were actually working with very similar sets of data. And both teams were analyzing that data. So business folks would find a problem. They would run an analysis in their tool chain to try to validate a hypothesis to solve it. And they would run over to the data science team and say, here, take this. A little bit like throwing this package over the fence. Please do x, y, and z. And as we all know, nobody likes having requirements or sets of data thrown over the fence. It's just not super productive. The other challenge with this setup of having our discipline separated was that each team had a very narrow focus. The business folks were trying to make impact in this quarter. The data science folks were looking a little bit further down the pipe, but nobody had really stepped back completely and said, oh my god. If we can really go deep into the size and fit problem, which, by the way, was introduced through the advent of industrial manufacturing, what could we disrupt? What could we change? How could we change the lives of our customers and really impact change in our business? So about almost a year and a half, two years ago, Zalando made a very big change in how we organized ourselves. We actually pulled together these cross-functional teams all centered around our customers' most important problems. So I was asked to lead this particular team. And the first thing I did was I said, OK, we've got all these different disciplines. Together, we need to build our strategy. So we took about six weeks, and we formed small cross-functional working groups, and we wrote strategy papers together. So what did this look like? I had a team that was looking at, what are the data sets that we have today? What are the data sets that we want to build tomorrow? And what kind of algorithms do we want to explore? And that team was not just comprised of data scientists. Actually had a business person in there who was able to challenge the data scientists to make sure that they could actually fully communicate their ideas. This kind of interdisciplinary sparring actually makes everybody better. And we created a very rich strategy that informs all the work we do going forward. The second thing we did was actually pull together around our immediate work. So we form initiatives every quarter that anyone in the team can pitch, and anyone in the team can work on. Now, we absolutely review these. A lot of them get thrown away. But the idea is you don't have to be coming from one discipline or another to actually pitch ideas about what we focus on. If you've got an idea, you can describe it. You can pitch it. The third thing we did, and this may have been one of the most important, was we actually adopted a framework to understand how we think about working with data. At the end of the day, the size and fit team is a data product development team. We all work with data, whether you're using the business intelligence stack to run analysis or a side kit and Python to actually start working with data and build models. We all work with the same information. And I pulled this particular model from a great book called Agile Data Product Development 2.0. It's from their manifesto. This is just a very simple way to think about working with data regardless of the toolset that you're working with. And the thing that this did was it really encouraged the business analysts who didn't have the advanced mathematics backgrounds to regard their work as just as important as the data scientist, really leveling the playing field for everybody in the team. What were the results? It all sounds great. But how do you actually measure success when you create a team that works this way? We grew by 25% this last year. And one of the things that I invested in was actually running a psychological safety survey to make sure that everybody in the team felt like they could ask questions without feeling like they were stupid or raise concerns when they saw something that might be problematic. It's a really great tool. I highly recommend it. There's a lot of material out there that you can go online and find. You don't have to design a survey yourself. It's a great way to understand how your team is functioning. The other thing that I'm happy to say is we actually were able to put a dent in the size-related return rates. This is super important for our customers as well as for our bottom line. And then the third thing that we achieved was we created this strategic vision. It's for driving towards 2020, which seems like right around the corner today, but at the beginning of 2017, this was a big deal. So in short, what I suggest you take with you, take away with you today, if you have an interdisciplinary team, really think about choosing a great framework to unify all the disciplines working in your team. Make sure that you as a leader, and we're all leaders when we play on a team, show respect for the value that each discipline brings. Take time and build that strategy together. It's really easy, especially when you're a startup, to kind of get caught up in the excitement of trying to execute and prove your value. You've got to do another round of fundraising, but without a strategy, all of those efforts and all of those activities, you can't really guarantee that they're going to give you the impact that you're looking for. So with that, I want to encourage you all. It's a lot more fun to work together. We learn a lot more when we work across discipline, and I think the only way to really achieve game-changing accomplishments, both for our businesses and for our customers, is when you work in these diverse backgrounds. So go forth and create great interdisciplinary teams. Thank you very much.