Our Recommendation
Precision Development (PxD) is a high‑rigor, highly aligned organization for our framework at the intermediate‑outcome level: it runs multiple RCTs, has a public experiment registry, and reports sizable, counterfactual income and adoption impacts for tens of millions of farmers.
Precision Development's Fierce Certification score is 90/100 based on our criteria:
✔ Has Ultimate Outcome Goals (50 pts)
✔ Measures Intermediate Outcomes (10 pts)
✔ Measures Ultimate Outcomes (0 pts)
✔ Shows Continual Learning & Adaptation (20 pts)
✔ Measures Intermediate Counterfactual (10 pts)
✔ Measures Ultimate Counterfactual (0 pts)
The Social Problem
PxD is targeting the persistent under‑performance of smallholder agriculture due to information, design, and delivery failures, which keeps millions of farmers in low‑productivity, low‑income traps. Farmers often lack timely, locally tailored advice on what to grow, how to manage pests and fertilizer, and how to respond to climate and price shocks, and public extension systems rarely incorporate behavioral insights or rigorous testing. As a result, recommended innovations are under‑adopted, incomes remain low and volatile, and families struggle to accumulate assets or escape poverty.
The Solution
PxD’s solution is to deliver cost‑effective, evidence‑based digital advisory services at massive scale, and to iteratively refine them using rigorous experiments. They send customised agronomic advice to farmers via low‑cost digital channels (SMS, IVR, apps) and increasingly AI‑enabled tools, grounded in local data, behavioural economics, and social‑learning theory. Their theory of change is that better‑timed, behaviorally optimized, locally relevant advice will increase adoption of profitable and climate‑smart practices, raise yields and income, and over time enable farmers to build more resilient, higher‑quality lives.
Key Outputs
Key outputs that contextualise PxD’s work:
- Scale: by 2024 PxD reached 18.5 million farmers (an 85% increase over the previous year) across Africa and Asia, and continues to expand.
- Experimentation infrastructure: the Experiment Registry documents more than 80 experiments, including RCTs and A/B tests, covering multiple services and countries.
- Evidence of cost‑effectiveness: PxD@10 describes that their digital agricultural services can deliver 12–19 dollars in additional income for farmers for every dollar invested in the service.
- Innovation beyond agriculture: projects like ElimuLeo in Kenya adapt PxD’s experimentation and adaptive‑learning approach to education, showing the transferability of their methodology.
These outputs underline PxD’s role as both service provider and evidence generator.
Key Intermediate Outcomes
Intermediate outcomes with strong evidence:
- Adoption of practices and input use: across evaluations, farmers offered PxD advisory are more likely to adopt recommended practices (e.g., improved seed, timely fertilizer application) and to adjust input use in more efficient ways than control farmers.
- Yield and profit increases: PxD’s RCTs show significant gains in yields and net income; aggregated across trials, they estimate 12–19 dollars in extra income for each dollar invested in digital advisory, with especially large benefits among poorer farmers.
- Service design improvements: adaptive experiments and A/B tests have shown that seemingly small changes (simpler menus, language alignment, different tones) can materially boost engagement and impact, and “simple tweaks that don’t lead to improved outcomes can be put aside.”
Key Ultimate Outcomes
PxD’s ultimate‑outcome evidence is still limited:
- Poverty status, food security, and resilience: public documents discuss improving farmer livelihoods and resilience, but do not yet present counterfactual measures of households moving above poverty lines, reducing hunger months, or achieving more stable consumption.
- Spending on education and health, and wellbeing: PxD emphasizes that higher incomes should enable spending on education and health and improve quality of life, but they have not yet reported RCT‑based indicators on these outcomes.
So, while PxD clearly aims at ultimate outcomes, the measured evidence remains at the intermediate level.
Continual Learning & Adaptation
PxD is notably strong on learning and adaptation:
- Integrated research and service delivery: PxD describes its research practice as intentionally embedded in service design, using experiments to both innovate and evaluate, and iteratively update its theory of change and learning objectives.
- Adaptive experimentation methods: their “Adaptive Experiments” and AI “missing middle” framework explicitly promote rapid, iterative testing (A/B tests, lab‑in‑the‑field experiments) before more expensive RCTs, to refine tools based on farmer feedback and contextual data.
- Open evidence sharing: the public Experiment Registry is an unusual commitment to transparency, enabling others to learn from PxD’s successes and failures and pushing the broader digital‑advisory field toward higher evidentiary standards.
In our four‑step cycle, PxD is a strong exemplar on grounding in a specific negative‑consequence chain for smallholders, designing interventions precisely aligned to that chain, measuring intermediate outcomes and counterfactuals rigorously, and feeding learning back into product and strategy.
Written by
