Theory of Change: #19: How to talk about failure

What happens when an organization built on maximum cost-effectiveness discovers that one of its flagship programs isn't delivering the results it promised?

For Evidence Action, the answer was clear — if uncomfortable: follow the data, own the mistakes, and scale back.

In a recent episode of Devex’s Theory of Change podcast, Evidence Action’s new CEO, Danielle Bayer, unpacked the nonprofit organization’s recent decision to wind down its Dispensers for Safe Water program across parts of East Africa, the backlash that followed, and what high-integrity implementation looks like in an evolving development landscape.

“Most organizations choose to make decisions and keep them internal,” Bayer said. “In this particular instance, we not only said that the program didn't work as we thought it should, but also that we made mistakes here.”

The decision to publicly air these operational mistakes sparked intense debate within the effective altruism and global development communities. While many praised the nonprofit's radical transparency, some critics questioned whether pointing to evidence was being used as a shield for poor implementation.

Bayer was candid about the trade-offs: “Admitting mistakes is hard, and it may mean you lose trust with some funders and the public. The truth is all organizations make mistakes. If we're not being transparent, it’s not that mistakes aren't happening — it's just that people aren't sharing them.”

She also spoke about what the future of cost-effectiveness looks like in the face of sharply declining global health aid, how Evidence Action codesigns programs with partner governments, and why she urges caution in the face of an upswell of interest in artificial intelligence.