Choice Exists, Even in Slums: How a Yelp-like model can benefit low-income communities
Right after graduate school, I moved to Guyana to work with the Ministry of Health. I visited many marginalized communities, including Issano, where I learned that pregnant women with complications were being sent on a five-hour boat ride to the nearest hospital while in excruciating pain. The central ministry, which designed the program, could’ve discovered such an issue by interacting with users of services and resolved it by providing appropriate training to the health workers in Issano. Yet systems are complex. The ministry hadn’t interacted with anyone in Issano for more than 10 years.
This was the first of many times in my international development career when I saw that organizations serving the poor rarely utilized user knowledge to improve their products and services. This frustration fueled the creation of TulaLens.
Our purpose at TulaLens is to enable low-income communities to provide feedback and retrieve crowdsourced information on the organizations that serve them. We thought, if organizations such as Yelp and Amazon can successfully crowdsource information on restaurants and clothing for upper- and middle-income communities, why can’t we do the same on health, education and other basic needs for low-income communities? Our ultimate aim is to help end users build a community through which they can help each other make more informed choices.
In our recent pilot, we tested the feasibility of this idea. We surveyed 114 pregnant women who lived in Jiyaguda and Mallapur, slum areas in Hyderabad, on their last prenatal care visit at a social enterprise, private or government facility. Some of the information we collected included wait time, time spent with provider and fees paid. We found that weak feedback loops were indeed a problem. Ninety-five percent of pregnant women had never been asked for feedback on quality of prenatal care services, 58 percent were only aware of one health facility in their area when 30 were accessible on average, and 100 percent found the data we shared useful. We then mapped out prenatal care facilities and the associated quality of care, and went back to the same women we originally surveyed to share the crowdsourced data.
Within one month, 25 percent of women had switched to a better facility, 43 percent had advocated for better services at their facility and 100 percent had discussed their facility choice with a family member.
Our pilot shows that choice exists in this urban low-income setting but is hidden by information asymmetry. This means crowdsourcing information can potentially be a powerful tool in these communities. Now, how do we go about expanding our reach?
We’re planning on growing slowly because there are many aspects of our approach that need to be refined, inclusive of four key areas. Through experimentation we’re hoping to start addressing some of the intricate challenges below.
1) Revenue generation: We’ve decided to provide TulaLens to users free of charge because they’re contributing time and expertise in exchange for quality data on organizations they use. User feedback also helps members of the community and organizations that serve these communities. Then how do we generate revenue to sustain ourselves? One possibility we’re exploring is to sell anonymized aggregate data to large companies who invest in NGOs and social enterprises. They can use this data to learn more about users in their area of interest and help current or future investees improve their products and services. Initially, we explored the possibility of selling data back to the organizations themselves. We quickly learned that small to mid-size organizations like the ones we encountered are often opposed to us talking to their users, and rarely have the appropriate systems – including human resources – in place to act on user data.
2) User base: Our approach has worked among a small group of pregnant women living in slum areas in Hyderabad who on average live on $1.44 (U.S.) a day. We want to deepen our expertise and impact low-income users of maternal and child health services more broadly before expanding to all health services. We’ll reach women both in Hyderabad and other cities in India. In the longer term, we’ll explore expansion to other sectors and other countries.
3) Method of communication: After testing several approaches and technologies, we believe users should provide and retrieve information in person the first time they encounter TulaLens. High response rates in each community are important because the crowdsourced data is location dependent. An in-person approach also builds trust with users and reduces sampling biases. We still haven’t found the optimal offline Android app for field researchers to collect and analyze data in person, and are considering developing our own. After the first encounter, we’ll experiment with existing systems so users can continue providing feedback and retrieving crowdsourced information via their basic mobile phone.
4) Content of feedback: We’re currently collecting a broad swath of information from users on demographics, quality of care and decision making. As we expand our reach we can use analytics to optimize the data we’re collecting. Shorter surveys are more efficient and can improve user experience.
Priya Iyer is the founder and CEO of TulaLens.