By Dante Donati, Victor Orozco, Ana Maria Munoz-Boudet and Nandan Rao
Over the past decade, internet access in low- and middle-income countries has surged, with 62 percent of the population online by 2022 (The World Bank). This growing connectivity has led governments, researchers, and development organizations to turn to social media for spreading important health messages. However, social media algorithms tend to prioritize the most active users, making it unclear whether campaigns reach and impact those who need them most—such as individuals at high risk of contracting malaria.
Evaluating the effectiveness of a social media campaign on real world outcomes
Our new World Bank working paper evaluated a malaria prevention online ad campaign in India. We addressed two key questions: did the campaign increase the likelihood that social media users and their families would adopt malaria prevention behaviors and seek medical attention when fever symptoms appeared? And did it work for those most vulnerable to malaria exposure? Our formative research revealed that dwelling type—solid structures versus non-solid dwellings—was the strongest household-level predictor of malaria risk.
Malaria No More, an international NGO, launched a six-month malaria prevention ad campaign across 22 states during 2020-2021. The ads were crafted for specific audience personas. To understand the campaign’s effectiveness, we conducted a study using a cluster-randomized control trial in three states: Uttar Pradesh, Chhattisgarh, and Jharkhand. We randomly assigned 80 districts to either a treatment or control group.
To measure behavior change, we collected data through chatbot surveys and health facility records from India’s Health Management Information System (HMIS)—an important innovation for objectively assessing the impact of a social media campaign in a developing country. We gathered data both during the campaign to capture immediate impacts and up to three months after its completion to assess the campaign’s cumulative effects. Our findings are based on intent-to-treat (ITT) effects, meaning individuals in treatment districts may not have seen the ads directly.
The campaign worked, but primarily for low-risk populations
We began by measuring the campaign’s impact while it was live, finding an overall 3 percentage point increase in bed net usage—a 4.5 percent rise compared to the control group. However, the impact differed by dwelling type. Among social media users in solid homes, bed net use increased by 9 –11 percent, while no significant effect was observed for users in non-solid dwellings. Those in solid homes were 13 percent more likely to seek medical care within 24 hours of developing a fever.
Data collected one to three months after the campaign ended shows that continued exposure had a measurable impact on malaria incidence. Among those in solid dwellings, self-reported malaria cases decreased by 44% at the extensive margin (whether any household member had malaria) and by 53 percent at the intensive margin (the proportion of household members that contracted malaria) compared to the control group. The reduction was strongest in households with existing bed nets, indicating the campaign reminded rather than prompted new purchases.
The HMIS health facility data supports our survey findings, showing a decrease in malaria incidence in urban areas, where solid housing is more common. Monthly malaria incidence fell by 6.2 cases per million people—equivalent to a 30 percent reduction from the pre-campaign monthly rate—over the nine months following the campaign’s launch. There was no significant change in malaria incidence in rural areas, where non-solid dwellings are more typical.
Did the ads fail to resonate with lower-income users, or did social media algorithms limit reach by prioritizing engagement?
The campaign could work for high-risk populations, if properly targeted
Follow-up surveys indicated that users in solid dwellings were more likely to recall the ads than those in non-solid dwellings. To further test this, we ran an individual-level feed experiment using Facebook’s Custom Audience tool, intensifying ad retargeting. Two weeks after this remarketing campaign, we conducted a follow-up survey. This time, the ad campaign improved outcomes equally across both groups, suggesting the nationwide campaign did not adequately reach those most at risk.
Social Media campaigns are cost effective
The social media ad campaign proved to be remarkably cost-effective. The campaign’s cost per malaria case averted (based on health facility data) ranged from US$3.40 in urban areas to $6.50 overall. Considering the true number of malaria cases could be over 20 times higher than reported, the cost per estimated averted case could drop below $0.20. This aligns with studies on online COVID-19 campaigns in the U.S., showing social media can be a scalable, cost-effective tool in developing countries for public health campaigns.
Policy implications
Our study highlights the power of digital platforms to deliver impactful public health messaging. However, it underscores the need for development partners to acknowledge the limitations of social media algorithms, which prioritize engagement over meaningful outcomes. Organizations must refine their targeting strategies using advanced tools like Facebook’s Lookalike Audience.
Development partners should consider allocating larger budgets to reach harder-to-engage, more vulnerable, and often higher-cost social media users, where the potential development impact can be greater. Measuring the effectiveness of social media campaigns should not rely solely on metrics like clicks or likes but on meaningful outcomes.
- The study was a joint effort between Facebook’s Campaigns for a Healthier World initiative, Upswell, Malaria No More, and the World Bank.
The post From clicks to care: Targeting social media to improve health in vulnerable communities appeared first on Caribbean News Global.
