Warehouse Picking Optimization to Improve Online Supermarket Efficiency

Warehouse Picking Optimization to Improve Online Supermarket Efficiency

Migros Online, a leading Swiss online supermarket and a part of the Migros Group, holds a significant position in Switzerland's grocery market. In recent years, Migros has become one of the key industry players, employing around 100,000 people.

To improve warehouse efficiency Migros Online initiated a project to implement a simulation-based digital twin into their system. This initiative led to a collaboration with Decision Lab, a company that specializes in decision intelligence.


Migros Online's main challenge was to improve how their warehouses worked, focusing on warehouse picking optimization. They wanted to implement a simulation-based digital twin to help their warehouse managers and IT team make better decisions.

They focused on two key areas to improve warehouse efficiency:

  1. Understanding and enhancing the warehouse layout, including where and how products are placed. Migros Online needed to find the best ways to optimize warehouse picking areas and routes that trolleys can take to move from one product to another.
  2. Upgrading Migros Online decision-making algorithms. This included figuring out the best way to pack products in shipping boxes, organizing picking trolleys, and assigning workers effectively.

The purple infographic with Migros Online business needs for this project

Migros Online business needs for this project

To optimize warehouse picking operations and overall improve warehouse efficiency, Migros Online worked with Decision Lab. They aimed to create a simulation model using AnyLogic software, a tool that would let them simulate and test different warehouse setups.


Why AnyLogic?

Decision Lab chose AnyLogic software for its ability to combine agent-based modeling to simulate the workforce and discrete event simulation for order processing.

The versatility of AnyLogic allowed the team to create a highly realistic model for warehouse picking optimization. For example, the model included a customizable transporter library. With this feature, Decision Lab simulated trolley movements and accurately represented product transportation within the warehouse to improve its efficiency.

Integration with external Java code was also a big part of why Decision Lab chose AnyLogic for this project. The software worked well with Migros Online's existing Java systems, which allowed it to interact smoothly with other Java-based tools.

Lastly, AnyLogic is a data-driven platform. This factor was especially important to accurately simulate the actual layout and operation of the warehouse. Decision Lab used detailed coordinates to create conveyors, turning CAD files into a workable AnyLogic model. This further allowed the warehouse managers to test different layouts and rules without changing the base code.

The infographic with reasons why AnyLogic was chosen for for this warehouse optimization project

Reasons why AnyLogic was chosen for this warehouse optimization project

The simulation model

The developed model showed how a Migros Online warehouse worked, right down to the details of how products were moved around.

One of the key parts of the model was its ability to handle different types of operations in the warehouse. For instance, it could show how workers do their jobs and how orders are processed.

Decision Lab built the model using real data, making it accurate. This meant that Migros Online could try out different ways of organizing their warehouse to improve its efficiency without changing the actual warehouse setup. The team also ensured that the model worked well with Migros Online's other Java-based systems, which was helpful for using the model in real-world situations.

The infographic with external Java optimizers

External Java optimizers

The AnyLogic simulation model gave a clear picture of how the warehouse works. It showed where products are stored, how much space there is for them, and the paths that trolleys take. The model even considered the specific needs of different types of containers and trolleys. Plus, it had a user-friendly interface with tools like heat maps (density maps) and information panels to help visualize warehouse activities.

The infographic with core components of the simulation model

The infographic with core components of the simulation model

Core components of the simulation model


Since Migros Online began working with Decision Lab, they have seen clear benefits in improving warehouse efficiency. The biggest advantage has been in planning changes to their fulfillment centers. With the simulation model implementation, they could test out new ideas for warehouse picking optimization before making any real-world changes. This was especially helpful for big, costly changes like rearranging the warehouse.

One of the early successes of this project was convincing Migros Online management to invest in improvements more easily. Migros Online could show Migros Group how changes would work through the simulations, making it easier to get the green light for actual upgrades. This led to a practical change that reduced their shipping costs by about 5%.

Looking ahead, Migros Online sees a lot of potential for implementing the simulation-based warehouse digital twin in other areas. As they build a new automated fulfillment center, they think this tool will help them work better with their automation providers. The digital twin should make it easier to show these providers where improvements can be made and to negotiate costs.

In the long run, Migros Online plans to use the simulation in other parts of their business, like delivery management. There's still plenty of room to make their current operations more efficient, and they're excited about continuing this work in the years ahead.

The case study was presented by Peter Riley from Decision Lab, and Wojciech Lapka from Migros Online at the AnyLogic Conference 2023.

The slides are available as a PDF.

Similar case studies

More case studies

Get a brochure with industry case studies