Part 10: Changes of perspective seldom happen in the comfort zone
Dr. Sabrina Zeplin, Director of Business Intelligence at the Otto Group in conversation with Uwe Weinreich
As Director of Business Intelligence, since 2012 Dr. Sabrina Zeplin has been responsible for developing the Otto Group into a leading player in the area of digital customer inspiration. Her division acts as an innovations laboratory and incubator for data-driven applications and business models for the Otto Group and its more than 100 subsidiaries. She has been at the Otto Group since 2007, and before her current role was the Divisional Director for Otto Group Consulting, and operationally responsible for the IT transformation of the subsidiaries in the German-speaking region of Europe.
Already published:
1. AI and Me – Diary of an experiment
2. Maths, Technology, Embarrassement
3. Learning in the deep blue sea - Azure
4. Experimenting to the Bitter End
5. The difficult path towards a webservice
7. Image Recognition and Surveillance
9. Seven Management Initiatives
10. Interview with Dr. Zeplin (Otto Group)
Uwe Weinreich: The Otto Group has extensive experience in the field of Business Intelligence, and, thanks to their history in mail order, possesses a treasure trove of data with quality and quantity that other companies can only dream of. Is what you and your team does the dream job for data analysts and Artificial Intelligence enthusiasts?
Dr. Sabrina Zeplin: Data scientists do indeed find an ideal environment with us: a great deal of interesting data, challenging applications and leading-edge technology. Technologically, we're constantly reinventing ourselves, and also developing Open Source solutions where previously there were none. Simultaneously, we're using the many exciting Open Source solutions which are arising in the global AI community, and developing them further. We believe we can solve problems in a completely new way, and thereby contribute to the excitement of our customers and achieve a lasting competitive advantage. To realise this, we must cover all bases: with an annual datathon, spread over several days, we've even institutionalised the experimentation a step further.
So you've successfully implemented not only Data Science, but also an agile working method. Which hurdles must today's companies clear in order to achieve this?
As well as the expertise in the subject, the answer lies in the company's culture. Am I prepared to share my data? Am I using existing synergies expediently? Do I realise in time when I need support, even possibly from another department or field? These and similar questions play a big role in times of digital transformation. We at the Otto Group are increasingly realising that new technologies can only be integrated productively when a wide portfolio of experts in their chosen fields are involved, which is why we're working ever more interdisciplinarily, in-among others-the area of voice commerce.
Was there something in this process which made you want to tear your hair out?
Almost every day, because we regularly make mistakes. We experiment, throw away what doesn't work, and try again. Sometimes, our initial problem was that we were too early to the party with some approaches. In one particular case we had a solution ready to go, but there was no interface in which we could have integrated it. In this case we simply didn't do our planning and research. What did we learn from this? Always plan thoroughly, then the implementation will be successful.
Doing this work interdisciplinarily, as you describe it, can also lead to friction between colleagues, can't it?
It would be fatal for the success of the business if we didn't do it this way! When for example, in the course of a technological project, software developers, data experts, marketers and strategists come together, admittedly they share a vision, but there's initially no common denominator. But then people start learning from one another and broadening their scope. Differences of opinion are also important for this process - as long as they remain constructive. Changes of perspective seldom happen in the comfort zone.
Does this change of perspective also lead to another kind of working and collaborating? And if so, what role does AI play in this?
I consider AI a strong driver for the whole area of digital transformation, because we've currently got more questions than answers regarding this subject. This is why we avoid committing ourselves to five-year plans. Digitisation is forcing us more and more to think in "sprints"- an approach we're familiar with from the area of agile working. I think this also explains the increasing relevance of Scrum, Extreme Programming and so on. Welcome change, and use mistakes as a basis for progress - that's the idea in theory, and increasingly also in practice. In our BI Group, for example, we hold regular Post-Mortem workshops, to be sure that we learn the right lessons from our failures. On a company-wide level, we also organise regular "Fuck-Up Nights", during which colleagues from all levels of the business reveal their biggest professional mistakes.
Do the results of this affect the strategy of the Otto Group?
Of course. But so do the results from other departments, as well as initiatives and project groups. Our aim is to become a fully digitalised trade and service company. It's becoming more and more important to convert knowledge and ideas from as many actors as possible into strategic decisions. Three years ago we initiated a cultural change in the company, and we are already seeing the benefits of cross-functional working: barriers are disappearing, and people who would never previously have worked together now work in cross-functional teams. And we are excited that more and more decisions are based on data or even completely automated.
And how do customers react to the change?
There are two answers to this. If we do our job well, the customer doesn't even realise that AI is involved. One example of this are the so-called "Aggregated Reviews", which we built for otto.de, and which are gradually being implemented by other companies as well. With the help of machine learning we've made a feature that's very important to customers, namely product reviews, still smarter, by generating summaries automatically. Thereby the user can search for reviews according to particular criteria. Value is front and centre here; the fact that a highly complex technological framework is behind it all is hardly recognised. We can see this by looking at the speed with which it's been adopted by customers. In a wider context, on the other hand, I feel the debate around AI is often driven by fear, especially when it comes to automisation. People are already talking about automisation as a job killer, even though there isn't really any evidence for this. And that's just one example. Here I think it's very important to offer clarification and to build trust - on the part of the company as well.
One more question: what advice would you give to managers who are just starting to experiment with AI?
My advice: get away from your desk more often, and visit the lab. You can only make good judgements if you try the technology yourself. It also helps immensely to look at things from the perspective of an end user. Basically, people should be the main consideration in all our decision-making.
Many thanks for the inspiring conversation!
published: January 7, 2019, © Uwe Weinreich