Background decoration Background decoration

Client case – Millions in discounts and less waste in packaging

Our customer is a leading company in biotechnology with operations in 180 countries. They package their products themselves into several tens of millions of final boxes. Then they ship them to patients all around the world. However, the supply planning for those packaging materials, notably cases, labels and notices, is a headache for several reasons.

The Challenge with Packaging

Firstly, the tightly regulated environment of pharmaceuticals means that the specifications to packaging differ government by government and change frequently. Those changes in requirements (“artwork changes”) are unpredictable and cause a lot of waste. It means that from one day to the next, all your inventory might be non-compliant and has to be scrapped. This represents millions of dollars in waste each year in addition to work against the sustainability goals of the company.

Secondly, economies of scale in the printing and packaging industry result in high discounts for ordering large quantities at a time.

In conclusion, you can either order large, discounted quantities which you have to throw away when an artwork change occurs. Or you can order many small and expensive quantities with a chance to avoid waste.

To mitigate this, our client currently defines a number of weeks to be covered by an order (“stockpiling”). This number has been calculated with a one-time ABC/XYZ analysis and gets more outdated by the day. Planners work on 2 different systems internally plus get spreadsheet reports every 2 weeks on upcoming artwork changes. Ordering the exactly right amount to minimize waste and maximize discounts is impossible in this environment.

Our approach

GenLots is a specialist in solving trade-offs to achieve a fully digital planning with the lowest total cost of ownership. Our client asked us to expand the scope of GenLots to packaging at one site. The aim is to solve the packaging challenge once and for all. In case of success, GenLots will be the digital standard which can easily be rolled out to almost 20 production sites around the world. Our approach has been the following:

  1. First, we added the packaging materials to our scope. The existing data exchange with the client made this easy.
  2. Then we established a business case with the GenLots simulator. We calculated the potential savings generated through our dynamic lot sizing approach.
  3. To minimize scrap, we worked out a way to define the maximum period an order should cover
  4. One of the requirements was that inventory will not increase with our optimization. We experimented and agreed on a weighting of inventory costs in our algorithm. This makes inventory levels stay the same as today.


Our proprietary reinforcement learning technology proved particularly fit to solve the packaging challenges. We balanced dynamically the objectives and got 10% of savings on total spend, mostly through a better use of discounts. The GenLots simulator recommended parametrization changes and won’t change the current way of working for this first step. This means over one million Euros of immediate savings on the test scope alone. The final scope of all sites would represent around 20x this number.

What’s next

We realized an immediate gain through reparameterization. The larger and longer lasting benefit will be to transform the process completely. We will determine lot sizes dynamically through GenLots. This will lower costs and CO2 tons emitted by reducing waste.

To achieve this, we initiated a project with the client to:

  • Write back the recommendations directly into the existing system
  • Integrate known artwork changes digitally. There will be no longer a need for manual spreadsheet transfers by planners. This avoids waste created by human error.
  • Roll GenLots out globally this year

We are looking forward to this exciting project and will keep you in the loop when we have managed the whole portfolio of several 100 million dollars of packaging material.

Client cases