A/B Testing SMS Messaging to Increase Clean Water Usage
Overview
To test whether SMS messages can influence household water consumption by delivering timely, persuasive messages to smartcard users. This is a clean, measurable experiment using existing smartcard infrastructure.
We want to understand what drives water usage. Rather than relying on assumptions, we are testing real-time behavioural responses to marketing messages. This will help us learn what works—and scale it if it does.
Why We’re Starting with Water
Water is the ideal starting point for this type of behavioural test because usage is already tracked digitally via smartcards. This gives us access to clean, real-time data. If messaging proves effective in shifting water behaviour, we will extend the approach to sanitation and hygiene, where measuring impact is more difficult but behaviour change is just as important.
Methodology
Population: 100 smartcard users who have consented to receive SMS messages
Design: Randomised controlled trial (A/B test) over 12 weeks
Intervention Group: Receives weekly SMS messages
Control Group: Receives no messages
Measurement: Monthly household water usage tracked via smartcards
Message Considerations
The content and scheduling of the messages (in Appendix 1) are shaped by several practical and behavioural factors:
Seasonality: The trial takes place during the dry season, when demand is highest and alternative sources are least reliable.
Multiple Use Cases: Messages focus not just on drinking water, but also on domestic uses like cooking, washing, and watering kitchen gardens—identified as core uses across nearly all households.
Survey Insights: Messages are grounded in recent data from the April 2025 Community Water Survey, which included over 100 structured questions. These revealed key usage patterns, barriers (e.g. cost, distance, taste), and household goals (e.g. feeding families, improving homes). Messaging content was tailored accordingly, including differentiated content for villages reporting poor water taste.
Barriers and Motivators: Messages are designed to pre-empt known concerns (e.g. bad taste, reliance on rainwater) and reinforce motivators like convenience, health protection, and the ability to meet daily needs.
Behavioural Framing: Each message includes a specific call to action. The aim is to prompt behaviour, not just awareness.
Language and Delivery: Messages are short, clear and in Kiswahili. They are delivered at times when users are more likely to act (early morning or late afternoon).
Data Integration: Because all recipients use smartcards, we can directly measure behaviour change through real usage data—making this a clean A/B test with no need for self-reporting.
Next Steps
Finalise message translations and delivery schedule
Launch campaign beginning July 2025, with weekly SMS delivery over three months
Analyse usage data in October to assess effectiveness
If results are positive, adapt and apply the messaging approach to sanitation and hygiene uptake
Why This Matters
This trial reflects DMDO’s commitment to practical, evidence-based innovation. A/B testing is standard in commercial marketing but rare in development. By grounding our messaging strategy in actual household behaviour, and rigorously testing what works, we are building a model for cost-effective, scalable behaviour change rooted in real-world data.
