Unlocking the potential of chatbots: Lessons from the SNEHA Didi pilot for urban informalsettlements

In the heart of Bhiwandi, Mumbai, a promising digital initiative is taking shape, aimed at revolutionizing how mothers in urban informal settlements access maternal and child health (MCH) information. The SNEHA Didi chatbot, a WhatsApp-based AI assistant, provides direct, timely, and reliable guidance to women juggling the challenges of motherhood in resource-constrained settings.

This tool goes beyond being just a technological solution; it aims to provide women with accessible and reliable information to make informed decisions about their health and their children's well-being. But how effective are chatbots in fostering meaningful engagement and addressing critical health needs? SNEHA's recent formative research sheds light on this question, offering valuable insights and recommendations for NGOs and development practitioners considering chatbots as a tool for community engagement.

For those interested in the technical aspects of SNEHA Didi’s development and deployment, check out the overview of the pilot project and insights into how it’s helping mothers embrace motherhood with confidence.

Key findings

1. Mothers seek help when problems arise
   Women primarily access MCH information during immediate crises, such as a sick child or feeding issues. Preventive health-seeking behaviours remain limited, underscoring the need for proactive interventions.
2. Trust in traditional sources remains strong
   Family members, healthcare workers, and doctors are the first ports of call for MCH information. While chatbots like SNEHA Didi rank as supplementary tools, they are yet to achieve the same trust level.
3. Engagement declines over time
   Despite initial interest, user engagement with the chatbot waned over the pilot phases. The need to refine content and interactivity to sustain long-term participation is evident.
4. Accessibility and convenience are key drivers
   Mothers appreciate the ease of accessing the chatbot from their homes, avoiding the time and effort of traveling to healthcare centres or the overwhelm of looking at multiple sources on the internet.

Lessons for Development Practitioners

1. Maintain trust and reliability
   Chatbots must deliver accurate, concise, and actionable advice. Regular updates and quality checks are essential to avoid misinformation or irrelevant responses.
2. Design for inclusivity
   Many users found menu-based interaction formats more accessible, particularly for those with lower literacy levels. Combining these with free-form questions ensures flexibility while reducing barriers to entry.
3. Leverage multimedia
   Video and voice messages resonate with mothers, especially those who struggle with reading. Offering content in multiple formats can enhance inclusivity and engagement.
4. Encourage family involvement
   Promoting chatbot use among family members can increase trust and adoption. SNEHA Didi's elder sister persona aligns with the cultural norm of seeking advice from trusted relatives.
5. Adapt to user preferences
   Timing matters - hours between 3 pm and 6 pm work best when mothers have free time.

Recommendations for Future Chatbot Initiatives

• Interactive Content: Incorporate quizzes and tutorials to make learning engaging and reduce dependency on frontline workers for onboarding.
• Proactive Messaging: Push regular messages to encourage preventive health behaviours, even in the absence of immediate concerns.
• User-Centric Design: Offer customizable formats, letting users choose between text, video, or voice messages.
• Scalable Solutions: Address common challenges, such as refining AI responses and integrating handovers to human advisors for complex queries.

To read more about the formative assessment and its findings, access the full report here.