Understanding the dynamics of the flow of goods

Colruyt laagste prijzen wanted to be able to cluster their branches and to gain insights on what levers can be used.

It’s surprising how such a model objectifies the essential factors that are necessary to move in the same direction.

Martine Pauwels, Director Logistics

Strategic Challenge

Colruyt laagste prijzen (CLP) wants to understand the dynamics of the flow of goods through their value chain. They want to be able to cluster their branches based on these dynamics and want to gain insights on what levers can be used to steer this in each cluster. In summary, the following questions need to be answered:

  • Why do some branches experience problems and others don’t?
  • Which characteristics should be used to steer these dynamics?


In-depth data analysis of the logistical characteristics/drivers

Möbius performed an in-depth analysis of the main logistical characteristics for the branches of CLP, including the forecasted and realized turnover, the physical surface areas of the branches and the delivery patterns. Typically, the data analysis for these drivers consisted out of:

  • The spread of branches across this characteristic
  • The identification, description and removal of outliers
  • The validation and interpretation of these data with the domain experts of CLP

Simulation of the flow of goods in the branches

Möbius developed a discrete event simulation model that describes the flow of goods in a CLP branch, influenced by its logistical characteristics.

  • Each branch was modelled with special attention to the flow of goods in their cross-dock zone.
  • Each logistical driver was parameterized, to study its effect and its interaction with others.
  • A data model allowed to link the real data from CLP’s systems. With this, a large simulation experiment (>22k simulated years, reported at an hourly basis) was set up to characterize the dynamics for all branches.

Statistical analysis and clustering of the simulation results

The results from the large simulation experiment were analyzed statistically to cluster CLP’s branches. This clustering was based on the observed dynamics of their flow of goods and the logistical drivers influencing this.

Möbius delivered a strategic clustering of the branches. For each cluster, Möbius formulated recommendations on which levers to work with to improve the flow of goods.


The approach has led to the following results:

  • Insights into the drivers for the dynamics of the flow of goods
  • Strategic model to cluster branches
  • Clear and differentiated recommendations per branch

Thanks for reading

Share case