$25 million saved while maintaining nutrition in WFP food assistance

Summary 

• Humanitarian food decisions are tightly connected: changing one food item can alter nutrition, procurement costs, transport routes and the number of people a programme can reach. 

• The team developed Optimus, a prescriptive analytics tool that starts from nutritional requirements and jointly optimises food baskets, sourcing, transport and the use of cash or vouchers. 

• In Iraq, the approach saved more than $25 million over two years; at the same funding level, the source study estimates that this amount could have covered food assistance for roughly 109,000 additional people each month. 

  
  

Why is food-assistance planning difficult?

The World Food Programme makes decisions across nutrition, procurement, transport and delivery. Historically, these choices were often handled separately: one team set the food basket, another planned how to move it, and another considered whether food, cash or vouchers were the most appropriate form of assistance.

But the decisions are connected. Replacing one commodity can change nutritional value, procurement markets, transport requirements and total cost. When funding is below the level of need, those trade-offs directly affect how many people can be reached and what nutritional value a programme can provide.

Why is this a prescriptive analytics problem?

The question is not only what will happen under a given plan, but what combination of decisions should be taken. Prescriptive analytics can search across many possible food baskets, suppliers, transport routes and delivery modes while enforcing nutritional and operational requirements.

That makes the problem a natural fit for mathematical optimisation: the model can work backwards from nutritional needs and identify a lower-cost way to meet them, rather than treating an existing food basket as fixed.

How did the partners build the solution?

Researchers and practitioners from the World Food Programme (WFP), Tilburg University, Northeastern University, UPS and the Amsterdam Business School co-developed one optimisation model covering nutritional requirements, food selection, local, regional and international sourcing, transport and the choice between food, cash and vouchers.

The model was developed into a web application called Optimus. It uses data WFP already collects and was refined through the operational knowledge of WFP teams working with food-assistance programmes in different contexts.

What do the numbers mean in practice?

In Iraq, Optimus identified a food basket that cost 12% less per month while providing virtually the same nutrition. Across two years, the reported savings exceeded $25 million. At the same funding level, that amount was equivalent to food assistance for roughly 109,000 additional people each month.

In Yemen, while WFP expanded operations from three million to six million people during conflict, the model made the trade-offs between funding, nutrition and reach explicit. This gave decision-makers a clearer basis for communicating what different funding levels could provide.

During the 2016-2017 El Niño drought in Southern Africa, the model also identified international sourcing options before local shortages and price increases intensified. Across these cases, the central impact is not a more complex plan; it is a clearer way to use limited resources while preserving nutritional requirements.

What can the findings mean for other applications?

The underlying approach can be applied wherever nutrition, sourcing and logistics decisions need to be made together. This could expand to large-scale feeding programmes, including school meals, as well as non-food humanitarian supplies such as health kits and mosquito nets.

Future work can also incorporate more uncertainty in prices, availability and demand, allowing the same prescriptive approach to adapt as operating conditions change.