Navigating chaotic times: forecasting amid the pandemic | Blog | bol.com

A massive amount of data

Originally from Poland, Eryk studied in Rotterdam, took a detour back to his home country, and once again returned to the Netherlands. “Before joining bol, I worked at a Fintech startup. It was a great experience, but one where I was wearing many different hats. I was eager to particularly focus on machine learning and I also wanted to join a more mature organization. That’s when bol caught my eye.”

Eryk continues: “Besides the maturity and the size, the possibilities in terms of tech and data at bol truly fascinated me. There is a massive amount of data available here, mostly well-documented, clean and well-maintained. With around 40 million items reaching millions of people in The Netherlands and Belgium, there is so much for me to work with. My job is to translate this data into forecasts for multiple use cases, like customer needs, logistics and products. To give you a practical example: with this kind of information bol can plan the staffing in our warehouses precisely, even 20 weeks ahead.”

Chaotic times

Joining bol in the fall of 2020, Eryk faced a particularly challenging period. He shares: “The pandemic completely disrupted bol´s existing forecasting models. We could no longer rely on past events and had to deal with a new situation without any control or available historical data. It was an interesting situation to be part of and one I had, obviously, never experienced before.”

Despite the challenging times, Eryk learned a great deal. “We organized regular brainstorming sessions, developed quick fixes, and pursued a long-term solution – all at the same time. Eventually, we incorporated the impact of COVID-19 using a tailor-made feature for our forecasting models, which successfully restored our accuracy levels. It was a really unique time to be part of, and it made me grow a lot as a professional.”