Unlimited room-for-more-funding on effective aid

By simongrinsted, Neal Lesh @ 2025-10-02T22:29 (+10)

tl;dr: Proven aid interventions fail to reach hundreds of millions of people who need them, in part due to delivery bottlenecks that create room-for-more-funding constraints. We see addressing these constraints as one of the most tractable opportunities in global health. In this article, we explore new approaches for scaling effective aid – through decentralization and standardization – and introduce a platform Dimagi is building for this: Connect.

--

For decades, we’ve known of cost-effective ways to address some of the leading causes of child mortality: insecticide-treated nets for malaria, ORS and chlorine tablets for diarrheal disease, and vitamin A supplements to prevent deficiency. Each of these interventions is inexpensive, evidence-backed, and proven to save lives. Yet collectively, malaria, diarrhea, and vitamin A deficiency still contribute to over a million deaths annually – most of them in children under five.

Part of the reason for this is straightforward: proven solutions aren’t reaching everyone who needs them. Only 68% of households in sub-Saharan Africa have a bed net; 47% of children with diarrhea receive ORS; and 59% of at-risk children receive vitamin A. These figures reflect real progress – driven by the efforts of evidence-minded governments, institutions, NGOs, and donors, including those in the EA community. But they also highlight the magnitude of the remaining coverage gaps. We believe bridging these gaps may be one of the most tractable opportunities in effective aid today.

Giving more consideration to how to scale proven interventions

To date, most EA funding for global health and development has supported scale through one of two strategies: expanding the capacity of the organizations already delivering an intervention, or building new ones from the ground up. These centralized approaches have some notable successes, but quickly run into two limitations:

  1. Operational constraints. Hiring and training new teams, building out internal systems, and adapting to new geographies, languages, and logistics – complex, resource-intensive processes that not all organizations have the capacity to do well.
  2. Diminishing fidelity. An organization implementing an intervention at a small scale faces fundamentally different challenges to the same organization implementing the same intervention at a large scale. Oversight becomes a key challenge, and with scale it often weakens. Delivery can diverge from the evidence-based model originally studied, and cost effectiveness can fall.

This can result in delivery bottlenecks: even when interventions are evidenced-backed and funding is available, the capacity to implement them well can be missing. Donors hit room-for-more-funding constraints and resources are redirected to less effective interventions, while large populations continue to go underserved.

Many ways to scale

These limitations are well recognized in EA and likely familiar to many readers. The primary goal of this article is to promote increased exploration of alternative approaches for scaling effective aid. Looking elsewhere, other parts of the social sector and the private sector offer a range of models for scale that could be worth exploring. Franchise models enable scale through many independent actors – relying on shared protocols, training, and quality assurance to maintain consistency. Platform models like Uber and Airbnb coordinate large networks of decentralized actors through digital infrastructure, with built-in mechanisms for verification, feedback, and accountability. These systems allow for service delivery that is locally responsive – adapted to different languages, norms, and contexts – while still maintaining fidelity to a common standard.

Most of these approaches achieve scale not by expanding a single organization, but by standardizing delivery and decentralizing implementation. We believe similar principles could be applied to scale proven-effective aid interventions – particularly the most operationally straightforward, such as distributing healthcare commodities.

Scaling through many actors

Part of the reason we’re sharing this here is to introduce Connect, an approach Dimagi is testing for decentralized scale in effective aid. Rather than scaling through a small number of large organizations, Connect scales through a large number of smaller, local organizations using a shared digital platform.

It aims to address both the constraints on centralized scale:

  1. It lifts operational constraints by partnering with locally-led organizations (LLOs) that already employ local staff, speak local languages, and understand the communities they serve. In a given region, we issue a call for expressions of interest from LLOs – such as NGOs, CBOs, or clinics already active in the area – to partner with us in delivering an intervention.
  2. It maintains fidelity through standardization, while allowing for context-specific adaptation. Frontline staff from partner organizations complete structured training via the Connect smartphone app, which then guides them through step-by-step workflows to deliver the intervention. Each delivery is verified with GPS coordinates and time-stamped photos, and partners are paid per verified service.

We applied this model in a GiveWell-funded project to use Connect for a Child Health Campaign (Connect-CHC) in Nigeria, deploying frontline staff of partner organizations to conduct door-to-door visits reaching 400,000 children with vitamin A supplementation and ORS with zinc. The intervention is straightforward and the commodities are low-cost, easy to transport, and durable. The challenge is effective distribution at scale. Connect approaches this through a repeatable process:

We have since scaled this model to seven other countries (CAR, DRC, Kenya, Liberia, Sierra Leone, Tanzania, Uganda) through an additional 18 LLOs. Collectively, these LLOs conducted over 775,000 door-to-door Connect-CHC visits in the first three quarters of 2025, with 665,000 in Q3 alone.  

We believe approaches like this – that focus on replicable systems rather than a single organization – can offer a more robust, cost-effective path to scaling effective aid.  

Scaling many things

We see potential for decentralized approaches like Connect to support a wide range of interventions. Our aim is to build a standardized delivery platform that can scale multiple evidence-based programs efficiently.

With support from GiveWell, Dimagi is currently developing another Connect program in addition to the child health campaign described above: a mother-baby wellness coaching program in which frontline staff are trained and paid to coach women in their community on maternal wellbeing and exclusive breastfeeding. We are also in various stages of developing programs for early childhood development, group therapy, kangaroo mother care, and distribution of non-prescription eyeglasses for reading. Each new program requires upfront investment in content development and configuration, but once built, can be deployed rapidly and repeatedly by many different organizations, anywhere in the world. 

We believe tech-enabled, decentralized, standardized systems like this can unlock significant room-for-more-funding for proven-effective interventions. In the long term, we envision these systems enabling hundreds of local organizations to manage thousands of frontline staff delivering dozens of effective interventions to millions of people.

Why we’re sharing this here

  1. To highlight what we believe to be a tractable challenge to the EA community and promote thinking around other ways to scale what we know works.
  2. To introduce Connect as one solution to this challenge. Dimagi is in the early stages of rolling out this platform, and we wanted to make others aware of the opportunity to scale effective programs this way – especially those who might be interested in partnering with us.
  3. To hear how this framing resonates with others thinking about effective aid, and open the idea up to feedback more broadly.