Title: Global Sales Forecasting | Brokk & Sindre Description: IKEA needed reliable sales forecasting across more than 30 countries with different market conditions, seasonal patterns, holidays and local demand. Canonical: https://brokk-sindre.dk/en/cases/ikea-forecasting - [Home](/en) - [Cases](/en/cases) - Ikea forecasting # Global Sales Forecasting IKEA needed reliable sales forecasting across more than 30 countries with different market conditions, seasonal patterns, holidays and local demand. By Mikkel Krogsholm · September 1, 2023 A model is only useful through its connection to the reality in which it will be used. Local seasons, weather and operations mattered more than an elegant algorithm alone. - Covered 30+ countries with local seasonal patterns - Forecasts for inventory and supply-chain planning ## Background As Senior Data Scientist II at [Teradata](https://www.teradata.com), I worked on sales forecasting for IKEA across more than 30 countries, from Sweden to Japan. ## My approach I built a forecasting pipeline that combined time series algorithms with local market understanding. The system accounted for everything from national holidays to local weather conditions. ### Technologies - **Python & R** for statistical modeling - **Spark** for distributed data processing - **Teradata** for data warehousing ## Results The forecasts were built for inventory and supply-chain planning across countries with very different local conditions. The work taught me how quickly a technically strong model becomes an organizational question about data, planning and accountable use. ## Related cases ### Crowd Flow Analysis for Smukfest When thousands of festival-goers switch stages at the same time, bottlenecks form. Without data, it's guesswork where the problems occur. ## Do you face a consequential AI decision? Write about the decision, or ask about a leadership briefing or talk.