Photo from livinggoods.org

Research

Living Goods

Todd Manwaring

on Sep 23, 2026

• 9 min read

Our Recommendation

Living Goods is one of a small handful of organizations in any sector that can point to randomized evidence on the outcome it exists to change. A 2019 trial randomized 214 villages across 10 Ugandan districts, evaluating a community health promoter program delivered by two organizations, Living Goods and BRAC Uganda, with roughly half the clusters in each. Pooled across both, under-five mortality fell 27 percent, infant mortality 33 percent, and neonatal mortality 27 percent after three years, at $68 per life-year saved. Broken out by implementer, the effect in Living Goods clusters was larger than in BRAC clusters, 31 percent against 23 percent Alongside that sits an independent evaluation showing unmet family planning need falling 36 percent against a control area, and a published quasi-experiment on how the organization pays its own workers.

Living Goods presents "two gold-standard research studies" showing a 27-plus percent reduction in child mortality. The second study's own abstract reports that across the full study period, under-five mortality did not significantly differ between intervention and control clusters. The 28 percent figure comes from one of two program variants during one of two phases, with the other variant showing no significant effect. That is still a meaningful finding, and the honest reading is that the model works when implemented well and does not work automatically at scale, which is arguably the more useful lesson. Living Goods links directly to the paper, so nothing is hidden, but the summary on the homepage is more confident than the evidence supports.

Two further notes. The 2019 trial evaluated an entrepreneurial sales-agent model that Living Goods has since largely replaced with government partnership, so the strongest evidence describes a predecessor of the current program. And the organization publishes its misses in unusual detail, including commodity stock at 21 percent against a 60 percent target in Burkina Faso and supervision at 43 percent against 60 percent in Vihiga. An organization that prints its own shortfalls in a donor report has earned the benefit of the doubt on the rest.

Living Goods' Fierce Certification score is 100/100 points based on our criteria:
✔ Has Ultimate Outcome Goals (20 pts)
✔ Measures Intermediate Outcomes (15 pts)
✔ Measures Ultimate Outcomes (20 pts)
✔ Measures Intermediate and/or Ultimate Counterfactual (20 pts)
✔ Shows Continual Learning & Adaptation (25 pts)

Donate Now! 100% Goes to Living Goods

Children in rural Kenya, Uganda, and Burkina Faso die of malaria, pneumonia, and diarrhea, conditions a trained worker with a rapid test and a course of treatment can resolve for a few dollars. Uganda's under-five mortality was 64 deaths per 1,000 live births at the time the first trial was designed, and the three conditions above accounted for more than half of them.

The failure is one of delivery, not knowledge. A child falls sick and nobody in the health system knows, because nobody is looking at household level. The nearest facility is hours away. If the family travels, the clinic may lack the medicine. The community health worker who might have bridged the gap is an unpaid volunteer with no supervision, no tools, and no stock. Meanwhile births happen at home, pregnancies go unmonitored, children go unvaccinated, and women who want to space births cannot.

Two further features shape Living Goods' approach. Data is invisible, so ministries cannot see where performance is breaking down or direct resources toward it. And outbreaks go undetected at community level for weeks, which the recent Ebola spread from the Democratic Republic of Congo into Uganda made concrete.

The Solution

The DESC standard. Every community health worker Living Goods supports is Digitized, Equipped, Supervised, and Compensated. That acronym is the organization's core claim: volunteer community health worker programs underperform because the workers are none of those four things, and fixing all four together is what converts the model from goodwill into results.

A three-pronged approach (What We Do). Learning sites, where Living Goods delivers services directly, tests innovations, and generates evidence. Implementation support, where it works side by side with governments to replicate what the learning sites proved, co-financing and co-implementing the DESC elements. And shaping the enabling environment, working on national policy, financing, and digitization.

Implementation Support 2.0, built on eight years of co-implementation with Kenyan counties, targets the five things that most often limit performance at scale: supervision quality, supervisor capacity, data use, co-financing for sustainable worker pay, and government management of digital tools.

The digital layer. Community health workers record interactions in national electronic Community Health Information Systems. Government Digital Champions handle first-line technical support. An issue-tracking platform called Afya Desk logs and resolves problems. Community Event-Based Surveillance workflows let workers report unusual health, climate, and environmental events into national dashboards.

Mapped against a Theory of Change, every component maps to a specific failure point named in the Social Problem Section, and the exit strategy, government ownership, is itself being piloted and measured. The one asterisk: the 2019 trial tested an earlier version in which community health promoters were locally recruited sales agents incentivized to sell health products. The current model is a government partnership. The second trial partly covers the transition, and its mixed result is consistent with a model that became harder to deliver as it scaled.

Key Outputs

Quarter 2 of 2026 (Q2 report):

12,959 active community health workers serving 6.2 million people across three countries
325,707 sick child treatments; 43,088 pregnancies supported; 54,526 couple years of protection; 13,293 unintended pregnancies averted
86 percent of children fully immunized; 95 percent of babies delivered at a facility
$1.24 net cost per capita annualized, ranging from $0.45 in Vihiga to $3.63 in Burkina Faso
516 community health workers and 89 government supervisors trained in Zorgho, Burkina Faso, where the government funds more than three quarters of program costs
2,650 digital issues resolved by government teams without escalation over six months across four Kenyan counties

Advocacy outputs the organization credits itself with contributing to: national budget allocations for community health promoter compensation and social health insurance premiums in Kenya, more than 6 percent of the national budget for health in Uganda, and more than one third of the funding needed for Burkina Faso's entire community health workforce.

Key Intermediate Outcomes

The intermediary layer is measured quarterly, by site, against published targets, with definitions attached. Very few organizations in this series report at this granularity, and fewer still print the numbers that missed.

Service coverage. Facility delivery at 95 percent against an 85 percent target. Four or more antenatal visits at 65 to 80 percent against 64 percent. Full immunization at 86 percent. Sick child referrals completed at 93 to 99 percent across most sites against a 68 percent target.

Service volume. 325,707 sick child treatments in a quarter. Under-five treatments per worker per month reached 32.5 in Burkina Faso learning sites and 25.8 in Uganda, both far above target.

System function. Commodity stock above 90 percent in Uganda. Supervision in the last month at 91 and 97 percent in the two learning sites. Government teams resolving 61 percent of digital issues without escalation.

Published shortfalls. Burkina Faso learning sites at 21 percent commodity stock against a 60 percent target. Vihiga at 43 percent supervision against 60 percent and 48 percent on-time postnatal care against 64 percent. Under-five treatments below target in Burkina Faso, Uganda, Busia, Vihiga, and Bungoma. The report names causes such as: amoxicillin stockouts, seasonal declines in malaria and pneumonia, and workflow gaps in the Kenyan application that cause family planning indicators to be underreported.

Two counterfactual results sit here. An independent evaluation of the Uganda family planning scale-up found unmet need down 36 percent, outperforming the control area by more than 10 percentage points. And a published multi-arm quasi-experiment on compensation found that fixed stipends combined with moderate performance-based incentives produced better health service delivery than the alternatives, evidence Living Goods generated about its own most contested design choice.

Key Ultimate Outcomes

Child survival is measured, and it is measured with the strongest design available.

The 2019 trial randomized 214 villages across 10 Ugandan districts. Treatment villages received at least one active community health promoter; control villages received none. After three years, under-five mortality fell 27 percent, infant mortality 33 percent, and neonatal mortality 27 percent, at $68 per life-year saved. The paper was published in the American Economic Journal: Applied Economics, registered with the American Economic Association, and profiled by the Abdul Latif Jameel Poverty Action Lab.

The second trial is the more important one for a donor, and it is more complicated. Conducted at scale in two phases, following 10,172 households to trial end with all-cause child mortality as the primary outcome, it found that across the full study period under-five mortality did not significantly differ between intervention and control clusters. During Phase II, one of two program variants showed a risk ratio of 0.72, 95% CI 0.54 to 0.97, and a 28 percent reduction. The trial is registered in the Pan African Clinical Trials Registry and remains a preprint.

The honest interpretation is that the model can reduce child mortality substantially and does not do so automatically. Which variant, delivered how well, matters. That is precisely the question the Implementation Support 2.0 work is built around, and it makes the organization's obsession with supervision quality and performance management look like a response to its own evidence.

Everything else on the ultimate outcome side is unmeasured. Maternal mortality, child nutritional status, disease incidence, women's agency over reproductive decisions, and outbreak severity. Immunization coverage is measured; measles cases are not. Malnutrition screening is delivered; growth is not tracked.

Both counterfactual layers are present, which is what distinguishes this organization from most in our directory. The ultimate outcome counterfactual is randomized. The intermediary counterfactuals cover family planning and worker compensation.

Continual Learning & Adaptation

A learning architecture. Living Goods' three-pronged model is literally built around evidence generation: learning sites exist to test innovations and generate evidence, implementation support exists to replicate what the learning sites proved, and the enabling environment work exists to get governments to adopt it. That is the four-step cycle drawn as an organizational chart.

Lessons travel, in both directions and across borders. Bungoma County's Implementation Support 2.0 results are being applied in Vihiga, Kisumu, and Busia, and a tailored version is being tested in Zorgho, Burkina Faso. Community Event-Based Surveillance was piloted in Kenya in 2025, the pilot surfaced specific lessons about verification delays, network connectivity barriers, and the need to define outbreak signals clearly, and Uganda is now scaling a refined version to nearly 50 districts. The Afya Desk issue-tracking platform was piloted in Kenya and is being adapted for Burkina Faso and Uganda. A cross-country learning session in Ouagadougou brought teams together explicitly for this purpose.

The model itself has been rebuilt. The organization evaluated in 2019 recruited sales agents to sell health products door to door and funded 60 percent of operations from product revenue. The organization today is a government performance partner whose stated goal is that governments own the systems. In Zorgho the government funds more than three quarters of program costs, and the district medical chief is quoted saying Implementation Support 2.0 is not a Living Goods project but a government program that Living Goods supports. That is the clearest evidence of Theory of Change revision available.

It researches its own operating decisions. The compensation quasi-experiment tested how to pay community health workers, which is the single most contested question in the field and the one where an organization has the most incentive not to look.

It publishes what went wrong and what it plans to do. The Q2 report names commodity stockouts, supervision gaps, application workflow problems causing underreported family planning data, and a specific performance improvement plan running August through December to address pregnancy registration and first-trimester antenatal care. It closes a transition of the Busia learning site into government-led implementation support and says so.

Learning shared externally. A peer-reviewed trial, a registered second trial, a research library spanning telemedicine, digital integrated community case management, family planning task-shifting, compensation structures, and community health worker financing, plus a commentary in The Lancet Primary Care on governing artificial intelligence in community health drawing on evidence from more than 18 countries.

Living Goods

Todd Manwaring

Our Recommendation

The Solution

Key Outputs

Key Intermediate Outcomes

Key Ultimate Outcomes

Continual Learning & Adaptation

Further Reading

Written by

Todd Manwaring

Our Recommendation

The Social Problem

The Solution

Key Outputs

Key Intermediate Outcomes

Key Ultimate Outcomes

Continual Learning & Adaptation

Further Reading

Be the first to know when a new organization is certified—and why it meets the standard.

Written by