Using primary data to enhance public service delivery

At DataDelta, we believe that social programs should be based on the needs of the people they are meant to serve. It sounds simple. But it can be hard to get reliable insights into the conditions and experiences of the billions of people who rely on these social programs. DataDelta is making it faster, easier, and more cost-effective to collect data directly from people so that their experiences shape the programs meant to serve them. In 2023 and 2024 DataDelta surveys represented hundreds of millions of people across India, Kenya and the Philippines. Next year we’ll represent over a billion in Asia and Africa. Let’s zoom in and see what this really means in practice.

In India, DataDelta partnered with the Telangana State government to assess access to health, nutrition, and education programs and measure outcomes on key social indicators. From a multi-sector survey of 50,000 households, we identified low levels of registration among pregnant women with Health Departments in their first trimester, missing essential prenatal services. The issue stemmed from the need to register separately with both the Health Department and the Women Development and Child Welfare (WDCW) Department. After the team identified this gap, the Health Secretary implemented integrated data sharing, streamlining the process. As a result, first-trimester pregnancy registrations with the Health Department increased twentyfold, improving access to early prenatal care and financial incentives for safe pregnancy, delivery, and newborn care. This simple data-driven fix has the potential to significantly boost health outcomes for mothers and babies across Telangana. 

In Kenya, the IDinsight data science and DataDelta teams worked with the Kenya Ministry of Health to predict household income for the most vulnerable families. Governments often need to do this to determine eligibility for social protection programs and contribution amounts for social services. This is a challenge as there are no formal employment records and socioeconomic data are scarce.   Our team improved the machine learning model that the Ministry was using to predict income by (1) diagnosing the sources of erroneous predictions, (2) recommending model changes, (3) collecting data from vulnerable households to correct the model’s errors, and (4) establishing an appeals mechanism for income classifications. Initially, the model accurately classified only 20% of low-income households. After our interventions, the model accurately classifies low-income families, ensuring inclusive and equitable access to healthcare services.

In the Philippines, the Department of Health’s Health Promotion Bureau used insights from DataDelta’s Health Promotion and Literacy Longitudinal Study to address significant gaps in health knowledge and practices. For example, the study revealed that only 1 in 5 Filipinos understood the vaccination schedule, leaving children at risk, and highlighted health literacy gaps among older, poorer, and less-educated individuals in rural areas. Many in these groups also lacked trust in the Bureau’s primary communication channel – social media. In response, the Bureau is increasing reliance on community health workers for direct citizen engagement, aiming to improve health systems and services for over 100 million Filipinos.

As these stories show, high-quality data is vital for equitable and inclusive social programs. DataDelta prioritizes data quality to preserve the integrity of survey responses. But maintaining quality comes at a cost. In 2024, we refined our methods to balance rigor and cost-effectiveness. We’re testing affordable alternatives to traditional sampling approaches, prioritizing the most impactful data quality checks, and using automated checks to target the more costly field-based checks. These efforts aim to make reliable, high-quality data more affordable and accessible to social sector leaders.

We are excited for 2025! We’ll be taking DataDelta to West Africa, deepening our work in the health sector in Kenya, helping Indian states understand and manage climate change, and helping local governments use data to improve social programs in the Philippines. We’ll continue using DataDelta for inclusive AI,  and gender equity and for working with partners to assess and improve the quality of the data they collect and rely on for decisions. We’d love to work with you, too, so please contact Sarah.Lucas@IDinsight.org if you want to collaborate with DataDelta.