Foundations have important but unrealized potential to contribute value to strategy by building, supporting, and engaging in learning. Their location in the landscape of social-change agents affords them a rare line of vision to see patterns and to work across boundaries of organizations, systems, and people. They have the resources to import both experience (from doers) and expertise (from researchers and experts). They can afford to experiment and try multiple solutions. If disciplined about their own commitments and biases, they can work to counteract the kind of blind spots and confirmatory tendencies that frequently undermine organizational capacity to learn.

Foundations have faltered, however, in maximizing this potential.

Foundations have downplayed the complexity of their work and in many cases ignored the uncertainties surrounding their strategies.

This article explores a series of self-created “traps” that hamper foundations in advancing the kind of robust learning needed to guide strategy in these complex environments, including:

1. linearity and certainty bias: when foundations frame their strategies as a set of linear, causal, and certain actions and fail to address the complexity surrounding the issues and systems they hope to change
2. the autopilot effect: when foundations distance themselves from strategy as it unfolds, thereby failing to learn from implementation
3. indicator blindness: when foundations track and monitor their strategies through performance indicators that reinforce the linear, causal, and often-unchecked as-sumptions built into the work.

Drawing on a deep body of work on action research and the authors’ experiences working in and with foundations, this article proposes a framework for avoiding these traps and advancing a practice of learning strong enough to validate their adopted role as strategists.