The data science market has reached a point that we can safely assume that if you have customers, data is valuable to you. It can help you find new customers, re-engage exisiting ones, and make product recommendations tailored to their specific needs. But for Westwing, an online style and furniture “magazine”, data can’t replace a good curatorial eye. We spoke to Global Head of Business Intelligence at Westwing Thomas Rützel, about fusing data and style to create the perfect product.
Tell us a little bit about Westwing.
Westwing is a leading international ecommerce company started in 2011 as the first online home and living shopping club in Germany. The idea behind our shopping club is to be a “shoppable magazine”: Like a magazine we inspire our over 19 million members with great ideas to make their home more beautiful. Additionally, our members can get the products that inspire them directly from us. We select the products with care, and offer them in curated sales campaigns of which we start several every day in each country. Thus, our website looks different every single day, which makes browsing inspiring and fun. Our members do not come to our website to buy a certain product. Rather, they come because they love to browse, because it is beautiful, curated with care, and because it is exciting – and we work hard to make this happen every single day.
We are now active in 15 countries across three continents, and generated EUR 111 million sales in 2013 – and this was only our second full year of business. This makes us one of the fastest growing ecommerce companies worldwide.
Complete our SAP x Data Natives CDO Club survey now, and help us to help you
What about your role within the company?
In principle, our role as Business Intelligence (BI) is generally to consolidate data, generate insights from it, and to share the data and the insights with the organization to improve our business. Two aspects make this role particularly exciting at Westwing: First, as you can imagine, we have tons of data, so there is huge potential to conduct insightful analyses. Second, Westwing’s culture is very data-driven. Hence, we not only have lots of data to mine and to understand, but also an organization that is very receptive to our insights: Instead of having to push hard to make sure our analyses have impact, our colleagues are actually always craving for more! Therefore this role is a lot of fun and very challenging at the same time: There are so many great ideas for analyses and there is an audience for each. We as BI have the ambition to shape Westwing with our insights!
Why are data science and business intelligence is so important to Westwing? Why do you think it’s so crucial that businesses in general adopt a data-driven approach?
Let me start with the second part of the question. The obvious benefit of adopting a data-driven approach is that this enables more fact-based decisions. When people have different ideas, often data is the best way to arrive at an objective decision. Adopting a data-driven approach can even open up entire new areas for business, and can unlock great competitive advantages. At the same time it is crucial to not only understand the great potential of a data-driven approach, but also its limitations: Not everything can or should be measured, and as BI we have to be very conscious about what part of reality we are measuring.
With this said, Westwing cultivates a unique combination of being data-driven and style-driven at the same time. Being data-driven means for us that data science and analytics are essential for us to understand our potential, measure our performance, identify areas we need to work on, and give indications on how we can do this. This opens up exciting opportunities for us: Just imagine for example the importance of rapid optimization of sales campaigns in our business model, the potential that lies in personalization, etc. – we can only address this potential with great business intelligence. Yet sometimes, we make a conscious decision in favor of style and inspiration even if data might point in the other direction. Consider for instance a classical AB test: While version A might perform better than version B in terms of conversion, we might still decide for version B simply because it looks more beautiful. This may sound unusual and counter-intuitive initially, but this way we can ensure that our site remains special and beautiful, something that cannot be measured easily.
Could you tell us a little bit more about specific use cases that you have come across, or specific problems you’ve managed to identify through BI?
One use case is the loyalty predictor we have built to understand how loyal our members are, the drivers underneath, and even predict future loyalty. This is essential for us because our business model is based on loyalty, meaning members coming to our website frequently. We have tested several predictive modeling techniques, and are now able to forecast loyalty with surprising accuracy. Since we identified the underlying drivers of loyalty, we are able to work with our colleagues from other departments to improve loyalty even further and to keep our members excited about Westwing.
A special and fun challenge was to measure and track the impact of TV advertisement in several countries. While this may sound easy if you have never done it, this is particularly difficult for a loyalty-based business model such as ours: We do not rely on customers coming to our site and immediately making a purchase, but attract members who are inspired by our site and visit us regularly and frequently. This means that the return on investment can be spread out over a longer period of time, and has to be measured against a baseline that is fast-growing even without TV. As no commercial solution fulfilled our needs of not measured only spot performance but also the overall impact of TV, we built our own TV tracking tool from scratch.
What are you working on at the moment?
To give just one example, we are re-working our data warehouse to leverage the potential of close to real-time data even more, and at the same time making more information available faster to larger parts of our organization. This is especially important in a business model such as ours, because our site, sales campaigns, and our products change every day. This requires many decisions by many people, and often we have to take them quickly. Since we have so much data to support these decisions, empowering more colleagues to find the data and analytics they need fast and easily is extremely valuable to us.
What are the advantages your customers get from you having this data-driven approach on the back end?
First of all our members benefit from having a greater website, with more inspiration and a better product offering. We can measure for example how well sales campaigns are received by our members, or how well they like the products we offer. This helps us to make our site even more exciting and interesting for our members.
A second advantage is that we use data to improve the order experience of our customers. This includes for example better and faster order delivery, and being more efficient in general helps us to be able to offer our products at exceptional prices.
When you’re hiring for your analytics team, are there any specific skills that you look for?
For the analytical part of the team a strong business sense is necessary. We work closely with Westwing’s other departments, and our analysts need to understand them well to be able to define the best measures to analyze and to interpret the results. Beyond this, and beyond the basics of strong quantitative skills, I look for skills such as predictive analytics and strong statistics. And, of course, any analyst who joins should get along well with our international and dynamic team!
What I like to see in addition is technical skills, such as the ability to write data import scripts in python or so, to round up the data scientist profile. For the technical side we are also looking for designated data warehouse developers who are effective in managing a large and complex data warehouse with many users.
What are your thoughts on the future of big data? How do you think it’s changing businesses?
I believe that, irrespective of the exact definition of big data, it will transform not just the way companies operate, but also entire industries. We can already see this in some leading industries already, when for example value generation is shifting towards utilizing large amounts of data fast, or when technology companies enter traditional manufacturing markets. I think that also companies who have not used data so much in the past will have to understand the role of data and its potential, because else it will become hard to keep a competitive edge in comparison to those companies who do.
All images courtesy of Westwing.