Keep the Water Flowing: The Hidden Crisis of Rural Water Management

Problem definition: In rural sub-Saharan Africa, people rely on communal hand pumps for clean drinking water. These hand pumps frequently break down and require repairs. Operating under the assumption that preventive maintenance is too expensive, local nongovernmental organizations (NGOs) often face difficult decisions about where to allocate their limited resources to reduce water point downtime. NGOs could invest in gathering more functionality information so that they can use their scarce resources on targeted water point repairs. Alternatively, they could conduct more preventive maintenance, increase their repair capabilities, or source more reliable and cost-effective spare parts to reduce repair demand. Methodology/results: To reduce water point downtime, we propose integrating preventive maintenance into existing programs. In collaboration with local NGOs, we conducted field research in Ethiopia and Malawi. We collected 47,240 observations of water point functionality from NGOs in Malawi, the Central African Republic, and Ethiopia. We then develop a Markov decision process model based on real-world practices to optimize maintenance schedules for NGO mechanics. We apply the model to field data from the three countries and find that incorporating preventive maintenance can reduce water point downtime by an average of 41.4% (ranging from 7.1% to 61.9%), often with little to no increase in logistics cost. We further conduct numerical experiments to examine the role of functionality information. For example, we find that, with high information availability, the reactive maintenance visitation approach is more effective only when repair demand is low. Managerial implications: Our findings challenge the belief that preventive maintenance is prohibitively expensive and advocate for its integration into NGO programs. We recommend that resource-constrained NGOs prioritize water point reliability, expand repair capacity, and reduce major repair costs over investing in extensive data collection. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2025.0005 .