Wheat farmers in Ethiopia achieve up to 38% higher yields with machine-learning-based fertilizer recommendations

Ethiopia is advancing from blanket fertilizer recommendations toward site- and season-specific nutrient management through the NextGen Decision Support Tool (DST), developed by the Alliance of Bioversity International and CIAT with the Ethiopian Institute of Agricultural Research (EIAR) and national partners. The tool integrates more than 50 years of agronomic and soil information, including over 40,000 crop-response trials and 20,000 soil profiles, with machine learning, crop modeling, geospatial data, and seasonal climate information to generate locally tailored fertilizer recommendations. Independent on-farm validation demonstrated wheat yield increases of 14–20% and additional net benefits of up to USD 665 per hectare compared with conventional recommendations, while broader piloting recorded yield gains of up to 38%. The advisory has expanded from initial pilots to 37 districts across five regions, reaching approximately 72,800 farmers, with about 10,000 farmers applying the recommendations and nitrogen-use efficiency improving by approximately 30%. Through the nationally coordinated HaFAS framework, the recommendations are delivered through public extension systems and digital channels, including mobile applications, interactive voice response, SMS, videos, chatbots, and private-sector platforms. The initiative demonstrates how harmonized data, advanced analytics, field validation, institutional partnerships, and diversified delivery mechanisms can support scalable, evidence-based fertilizer advisory services that improve productivity, profitability, resource-use efficiency, and climate resilience among smallholder farmers.