Driving Behavior Change at Scale: Lessons from African Impact Programs

Behaviour change is a listening problem first

Across 75+ programmes in Africa and Asia, the pattern is consistent: campaigns fail less often because the message was wrong and more often because the hesitancy driver was never identified. Risk Communication and Community Engagement (RCCE) works when monitoring comes before messaging — when teams can see which concerns are spreading, where, and among whom.

Localised AI models make that listening practical at national scale. Sentiment monitoring across digital, radio and volunteer-reported channels surfaces resistance hotspots early enough to respond with correction rather than crisis management.

Trusted voices outperform reach

The single most transferable lesson from our health programmes is that credibility beats audience size. In Kenya's informal settlements, hyperlocal key opinion leaders helped a COVID-19 vaccination campaign reach 469,000 people against a 94,000 target. In Togo and Côte d'Ivoire, peer engagement around DMPA-SC self-injection produced over 113,000 engagements and a 61% conversion rate.

These are not celebrity numbers; they are community numbers. Health workers, religious leaders, agri-influencers and peers convert because the audience already trusts them on the specific topic at hand.

Design for measurement from day one

Impact claims only hold up when the measurement plan exists before the campaign launches. We combine baseline and endline surveys with engagement analytics and outcome tracking, and use randomised or quasi-experimental designs where the programme allows, so that reported change can be attributed rather than asserted.

In the AMREF RCCE programme across Tanzania, Uganda and Kenya, 900 baseline and endline survey responses sat alongside a campaign that reached 11.2 million people — the surveys are what turn reach into evidence of behaviour change.

• Define the behaviour, not just the message.
• Instrument baseline before the first post goes live.
• Align indicators to SDG targets so results travel across funders.

Emergency response needs a standing capability

Outbreak response cannot be assembled from scratch. Programmes that can activate within 72 hours do so because monitoring, message-testing loops and community networks already exist and are simply re-pointed at the new emergency.

Funders increasingly value that standing capability over one-off campaign delivery, because it converts preparedness spend into response speed when it matters.