Médecins Sans Frontières (MSF), Analytics for a Better World and ORTEC developed a simulation model to test how changes in distribution, stock levels and replenishment could affect MSF’s supply network before those changes are made in practice. The model is not yet being used operationally, but MSF has already tested whether it can support priority questions around regional distribution centres, inventory and supply reliability.
Why does MSF need to rethink how supplies move through its network?
MSF’s supply network moves far more than medicines. It also includes laboratory materials, hospital equipment, communications equipment, spare parts and other goods needed to keep medical operations running.
Historically, MSF’s operational centres have had a high degree of autonomy. That autonomy is important because medical needs can change quickly and teams need to respond locally. But it also means different parts of the organisation can maintain separate procurement routes, suppliers and stocks for the same countries.
An organisation-wide study of MSF’s supply footprint identified opportunities to reduce stock, cost and delivery times by moving more products through regional distribution centres rather than sending many bulky, frequently used items through global centres in Europe. MSF is now expanding regional distribution capacity in Kenya and considering a new centre in India.
The challenge is deciding what should move through those centres, how much stock each location should hold and whether additional hubs would improve the network.
Why is this a prescriptive analytics problem?
These are decisions about what to do, rather than simply questions about what has happened.
For example:
Which products should be stocked regionally?
How much inventory should sit at global, regional and country level?
Would another distribution centre reduce costs or delivery times?
When should products travel by sea, and when is faster air freight necessary?
A simulation model allows MSF to test different versions of the supply network without changing the real system first. It can estimate how alternative policies affect measures such as inventory, transport, delivery time and the risk of products being unavailable when needed.
One future application is multi-echelon inventory optimisation. In plain language, this means deciding how much stock should be held at each level of a supply network — for example at a global warehouse, a regional distribution centre and closer to a medical project — rather than treating each warehouse separately.
MSF is initially concentrating on its more predictable supply flows, which account for around 90% of its work and expenditure, while emergency supply remains a separate area with dedicated stocks and capacity.
How did MSF, Analytics for a Better World and ORTEC develop the model?
The collaboration began with the operational questions rather than the technical model.
MSF and the analytics team discussed what the organisation was trying to improve and which supply policies were worth testing. Johan van Schagen, Business Analyst for Supply, describes those requirement-setting conversations as one of the most valuable parts of the collaboration. Rather than simply implementing a requested policy, the technical team asked what MSF was trying to optimise and proposed alternative approaches that could also be tested.
The model was also designed with future integration in mind. MSF’s reporting and analytics team was involved early so that the eventual solution could fit the organisation’s existing data infrastructure and also be run locally for testing.
This matters because the aim is not to produce a one-off analysis. The model is intended to become a reusable environment for comparing different supply-chain decisions as MSF’s regional distribution strategy develops.
What questions can the model test?
One immediate area is regional distribution.
MSF already operates a distribution centre in Kenya and is preparing to expand its role, while a centre in India is being considered from the ground up. Simulation could allow the organisation to compare different network designs before investing in additional warehouse capacity or changing supply routes.
A second area is inventory. Once products are routed through both global and regional centres, MSF needs to decide how much stock should be kept at each location.
A third is replenishment. Much of the regular supply network works around relatively long ordering cycles, including time for processing, picking, packing, customs clearance and international transport. MSF is also seeking to consolidate freight into sea containers where possible and use air freight when necessary.
Simulation gives the organisation a way to compare these choices before applying them across a network that supplies medical projects in many countries.
Why does better inventory planning matter for medical operations?
A supply-chain model is useful only if the measures it improves connect to what medical teams need.
One particularly important measure for MSF is a stock rupture. A missing item does not automatically mean a medical activity stops. Teams may source an alternative locally or substitute another product, but those responses take time and may not always be possible.
MSF therefore wants future analysis to distinguish between products according to their medical importance. A life-saving medicine, for example, should not be treated in the same way as a non-critical item simply because both appear in the same supply system.
That requires input from medical, operational and supply colleagues because the importance of an item can also vary by context. The eventual value of the model is therefore not simply lower inventory or transport costs. It is the ability to test whether resources can be allocated differently while maintaining reliable access to the products that medical teams consider most critical.
Why is change management part of the challenge?
Developing the simulation model is only one part of changing how MSF manages its supply network. Using it in practice requires agreement on which decisions should be standardised, which should remain local, who needs to be involved and which projects should be prioritised first.
That is particularly important in MSF because operational autonomy is central to how the organisation responds to rapidly changing medical needs. Country programmes and operational centres need room to act quickly, while a more integrated supply network requires greater coordination across procurement, logistics and inventory. The challenge is therefore to introduce more structure without removing the flexibility that emergency and medical teams depend on.
The model also crosses organisational boundaries. Decisions about stock do not sit only with supply teams: medical colleagues need to define which products are most critical, while operational teams may need to change forecasting or ordering practices. That means implementation requires input and commitment across supply, medical and operational functions rather than ownership by one analytics team alone.
MSF did not describe the main challenge as a lack of support for analytics. The harder question is what to prioritise first. There are many strategic projects competing for the same organisational capacity, so introducing the simulation model means deciding which use case should move into practice first and which other work may need to wait. MSF needs to decide which use case to prioritise, which teams need to be involved and how simulation outputs will fit into existing supply, medical and operational decisions.
What has changed so far?
The simulation model has reached a stage where MSF believes its core capabilities can support the use cases identified. The remaining challenge is to connect it to a specific operational priority and build the governance, workflows and ownership needed for teams to use it in practice.
The impact at this stage is a decision-support capability that MSF can draw on as specific supply projects move into implementation, together with a clearer understanding of which questions the model can address.
What can this mean for MSF’s supply network next?
The next step is to connect the model to one of MSF’s strategic supply priorities and use it in an operational decision.
Distribution-network design is one possible starting point: MSF could compare different configurations for regional hubs before committing to infrastructure or changing flows. Inventory policies are another, particularly where better stock planning could reduce both excess inventory and the risk that critical products are unavailable.
Data quality will develop alongside that process. MSF currently has gaps in some country-level consumption data, but the organisation does not want to wait for perfect data before improving forecasting and inventory practices. Introducing more structured policies can itself create stronger incentives and processes for improving the underlying data.
The next phase is to determine which supply decision to test first, integrate the model into that process and measure whether the resulting changes improve cost, lead times, stock availability or other operational outcomes.
