A warehouse planner at the Zambia Medicines and Medical Supplies Agency (ZAMMSA) faced a familiar challenge in public health logistics: sequencing medicine deliveries across hundreds of geographically dispersed facilities without the data systems to do it reliably. Orders, facility-level consumption records, and service delivery data each lived in separate silos. Without a tool to integrate and optimize across those variables, planners worked from what information was available to them, and schedule adherence, or how often medicines actually reached health facilities when scheduled to, sat at around 50 percent.
Half the time, the plan held. Half the time, it didn’t. For a public health supply chain serving hundreds of facilities across a geographically dispersed country, closing that gap required something more than effort or expertise — it required a purpose-built optimization system.
ZAMMSA and Chemonics International built one.
The Tool
The Dispatch Optimization Tool, or DOT, combines commodity volumetric data with facility-level orders and GIS information, then runs the result through mathematical optimization algorithms on open-source platforms: OpenRoute Service, Google OR-Tools, and Python. The output is a concrete dispatch plan outlining which facilities to group into a single route, what order to visit them, and which trucks to assign to which routes.
The interface was designed for warehouse staff across a range of technical backgrounds, making it accessible to planners who manage day-to-day logistics without data science expertise. When circumstances change, routes can be replanned rather than rebuilt from scratch.
The outcomes were specific. Distribution adherence increased to approximately 99 percent as deliveries reached more than 650 health facilities on schedule. Warehouse coordination improved, and planning became more responsive to real-time conditions.
Why Last-Mile Distribution Is Hard
The phrase “last-mile delivery” describes an operational problem that consumes substantial time and routinely breaks down in ways that can be difficult to track across dispersed facility networks. In low-resource settings with limited road infrastructure, the complexity compounds.
The underlying challenge is data fragmentation. Health supply chains are only as good as the information flowing through them, and last-mile visibility on commodity availability and consumption remains elusive in systems where orders, facility-level consumption records, and service delivery data each live in separate silos. Even where central procurement is functioning well, warehouses cannot easily make good dispatch decisions without reliable data on what was consumed, what’s running low, and where.
What DOT addresses is a narrower slice of this problem: routing and scheduling. ZAMMSA’s procurement decisions are handled upstream. The breakdown is in sequencing deliveries across a dispersed facility network so that what’s already in the warehouse actually arrives on schedule. That’s a bounded, solvable problem. The 50-to-99-percent jump shows what solving it yields.
The Broader Context
The Zambia program sat within a much larger operational context. Chemonics’ work in supply chain management has supported procurement and delivery of essential health commodities to 93 countries, including antiretrovirals, malaria treatments, maternal health supplies, and emergency commodities.
Managing over 5,600 items from more than 410 suppliers is a complex catalog challenge. Automation is what makes it traceable. Chemonics has indexed and automated commodity catalogs across this work, applying the same operational logic that DOT uses for distribution routing: replace manual management of complex variables with a system that handles the optimization, and direct human judgment toward decisions that genuinely require it.
Building for Handoff
The DOT was created to run on open-source software. The choice was deliberate. Tools built on proprietary platforms often become stranded when a contract ends or a vendor relationship changes. Dynamic routing tools simplify the many variables that factor into last-mile distribution, but the transfer of that capability to national agencies is a key component of what the model is designed to achieve. Open-source builds give host-country agencies the ability to modify, maintain, and expand the system without depending on external technical support indefinitely.
Chemonics builds toward handoff: the point where the tool belongs entirely to the health system it serves.
A program that delivers 99 percent schedule adherence for one contract cycle is useful. A tool a national medicines agency can run, adapt, and own long after the contract closes is something different.