For the second year in a row, Zeenea participated at Big Data Paris as a sponsor this past 11th and 12th of March to present its’ data catalog.
During the event, we were able to attend to many different conferences presented by professionals in the data field : chief data officers, business analysts, data science managers, etc…
Among those conferences, we had the opportunity to attend the Zalando conference, presented by Kshitij Kumar, VP Data Infrastructure.
Zalando: the biggest eCommerce plateform in Europe
With more than 2,000 different brands and 300,000 items available, the German online fashion platform conquered 24 million active users in 17 European countries since its’ creation in 2008 .
In 2018, Zalando earned about € 5,4 billion : a 20% increase since the year 2017 !
With these positive results, Zalando has a lot of hope for the future. Their objective is to become the fashion reference :
“We want to become an essential element to the lives of our customers. Only a handful of apps make it to being part of a customer’s life such as Netflix for television or Spotify for music. We aim to be this one fashion destination where the customer can fulfil all of their fashion needs. ”
explains David Schneider, co-CEO of Zalando.
But how was Zalando able to become so successful in such little time? According to Kshitij Kumar, it is a question of data.
Zalando on the importance of being a data-driven enterprise
“Everything is based on data.” states Kshitij Kumar during his conference Big Data Paris this past March. For 20 minutes, he explains that everything must revolve around data : business intelligence and machine learning are built based on the company’s data.
With more than 2,000 technical employees, Zalando claims a Big Data infrastructure in different categories :
In response to the GDPR, the VP Data Infrastructure explains the importance of establishing data governance with the help of a data catalog: “It is essential to an organization in order to have safe and secure data.”
A machine learning platform
It’s by exploring, working, curating and observing your data that a machine learning platform can be efficient.
It’s by putting into place visual KPIs and trusted datasets that BI can be proactive.
Zalando’s Machine Learning evolution
Kshitjif reminds us that with Machine Learning, it is possible to collect data in real time.
In the online fashion industry, there are many use-cases: size recommendation, search experience, discounts, delivery time, etc…
Interesting questions were then brought up: How can you know exactly what a customer’s taste is? How to know exactly what he could want?
Kumar answers by telling us that it’s by repeatedly testing your data:
“Data needs to be first explored, then trained, deployed and monitored in order for it to be qualified. The most important step is the monitoring process. If it is not successful, then you must start the machine learning process again until it is.”
Another benefit in Zalando’s data strategy is their return policy. Customers have 100 days to send their items back. Thanks to these returns, Zalando can gather data and therefore, better target their clients.
Kshitij Kumar tells us that by 2020, he hopes to have an evolved data structure. “
In 2020, I envision Zalando to have a software or program that allows any user to be able to search, identify and understand data. The first step in being able to centralize your data is by having a data catalog for example. With this, our data community can grow through internal and external (vendors) communication.”
 “L’allemand Zalando veut habiller l’Europe – JDD.” 18 oct.. 2018, https://www.lejdd.fr/Economie/lallemand-zalando-veuthabiller-leurope-3779498.
 “Zalando veut devenir la référence dans le domaine de la mode ….” 1 mars. 2019, http://www.gondola.be/fr/news/non-food/zalando-veut-devenir-la-reference-dans-le-domaine-de-la-mode.
 “Zalando Back in Style as It Bids to Be Netflix of Fashion – The New ….” 28 févr.. 2019, https://www.nytimes.com/reuters/2019/02/28/business/28reuters-zalando-results.html.