Evidence in the Room, Change Out the Door: When Data Fails to Drive Decisions

During the Mpox emergency across six main African countries, UNICEF and partners generated one of the richest streams of behavioural and community insights seen outside COVID-19: community rapid assessments, qualitative studies, digital listening, and feedback loops ran continuously. Yet programming barely shifted. IEC posters, generic hygiene messages and centralized surveillance persisted, even as evidence showed transmission through close contact, cramped living conditions, stigmatized groups going underground, contact tracing missing most cases, and gender-blind spots in vaccine rollout. This talk unpacks why high-quality insights failed to translate into action: expert hierarchies that discount community evidence, centralized planning far from affected people, path-dependent budgets, and an activity-over-outcomes culture. It also spotlights counterexamples (e.g., rapid RCCE-driven pivots during Zambia's cholera response) to show that insight-to-action is possible with political cover, cross-pillar teamwork, and action-tracking. We propose practical fixes: make social/behavioural evidence structural (in data and decision cycles), fund adaptation instead of replication, track the use of evidence and outcomes, standardize community co-decision, and seed champions beyond SBC. The goal is not more data; it's different decisions. Our failure to act on what people told us cost time, trust, and equity. Unfortunately, this is a pattern across all epidemics. Let's make "using evidence about people" non-negotiable in future responses.
We will present examples of compelling evidence, data, charts, and build the story around how these insights should have shaped our actions and why they were ignored.

Despite ambitious investment and rigorous evaluation, Save the Children's INSPIRING trial in Jigawa, Nigeria (2020–2022) failed to achieve its primary outcome of reducing child mortality. The trial tested a participatory whole-systems model-men's and women's groups, Partnership Defined Quality Scorecards, and health worker training-implemented across 32 clusters with over 12,000 children. The hazard ratio for mortality showed no significant difference between intervention and control groups (HR 0.95, 95% CI 0.68–1.33; p=0.79).
While this outcome was disappointing, it created space for honest reflection. The process evaluation revealed mechanisms of change-improved women's wellbeing, increased pneumonia knowledge, trust in health workers, and social norm shifts-but these changes were not yet strong or widespread enough to translate into measurable mortality impact within two years. Challenges included incomplete coverage of the women's and men's groups, COVID and flooding that disrupted implementation and data collection, and uneven institutionalisation of quality improvement approaches.
By sharing this experience, we aim to normalise open discussion of failure in our SBC interventions – namely the insufficient coverage of our key intervention - men's and women's groups. Our lessons show the importance of adequate intervention dosage, deeper consideration of gendered household dynamics, stronger government ownership, flexible design to respond to shocks, and sufficient time for social and behaviour change to translate into survival gains.

The PEBIC framework, a monitoring tool that assesses Presentation, Empathy, Bias elimination, Information Correctness, and Context Specificity. It was deployed to evaluate live radio programs across several Nigerian states during a family planning project.
The objective was to ensure the quality, accuracy, and consistency of messages disseminated to the public.
According to the criteria, the findings showed that resource persons delivering the live programs in certain regions scored significantly low scores on empathy and bias elimination. For instance, one resource person humorously remarked that "family planning works 99%, and the remaining 1% is for God because only He is perfect," instead of providing a clear and factual explanation, such as family planning methods are highly effective when used correctly, but no method other than abstinence is 100% guaranteed.
Another resource person advised a male caller that "if his wife wants more children, it's her choice; he can only advise because the country is hard," rather than offering a balanced response, such as family planning decisions should be made jointly by couples. These scenarios showed the importance of consistent monitoring to ensure that accurate messages were communicated with empathy.
A closer look revealed that humour, cultural idioms, sarcasm, and religious framing were normal and effective ways to connect with the listeners in those regions, this was flagged as inadequate by evaluators.
 The learning here is that we cannot measure communication by uniform standards but by how well messages resonate within local realities.

Despite progress in antenatal care (ANC) service quality, maternal mortality remains high in India. Many causes stem from both systemic constraints and gaps in caregiver behavior. Noora Health's Care Companion Program (CCP) aims to address these by educating pregnant women and their families on essential ANC topics such as nutrition, warning signs, birth preparedness, and early breastfeeding, through structured, engaging sessions led by nurses in public health facilities.
However, CCP's ANC program faced early challenges. Compared to postnatal modules, ANC sessions showed lower attendance, retention, and practice uptake. The outpatient department (OPD) environment - crowded, transient, and transactional - was not conducive to group learning. Women often arrived fatigued or anxious to leave quickly after check-ups, making in-depth sessions impractical.
This session explores how our initial "failure" in implementing ANC CCP led to an intentional redesign grounded in HCD principles. Through observation, interviews, and co-creation with nurses and pregnant women, we learned that effective ANC education must mirror the lived realities of busy outpatient settings - short, modular, easy to recall, and integrated into routine waiting times.
Our reflections reveal how embracing failure as feedback can strengthen program design, shifting from "delivery of content" to "facilitation of experiences." We discuss practical lessons on redesigning health education to work with the system, not despite it.