Organizations should stake a balance between type 1 decisions, which are big, significant and risky; and type 2 decisions, which are low-risk, reversible, and offer new insights. At Amazon, for example, small teams are empowered to make decisions and act on type 2 decisions without heavy bureaucratic control; this helps ensure Amazon has a steady stream of new insights.
Many organizations make significant, type-1 decisions based on large assumptions that may or may not hold true. But as information comes back and some of these assumptions are called into question, they succumb to the sunk cost fallacy: they’re unwilling to concede the considerable capital they’ve expended on their idea and so double-down on their efforts until it is far too late to save the project.
A better approach is to de-risk by making smaller, type-2 bets on an ongoing basis and learning from the results. Investing in multiple, inexpensive ideas in parallel helps the organization invalidate a greater number of ideas more quickly. Each iteration makes the product or solution stronger by eliminating non-viable options.
This approach also lowers the stakes of our work. By treating ideas as experiments to be tested, we grow accustomed to the notion that they're working hypotheses rather than solutions.
Relatedly, Richard Rumelt advocates the pursuit of "proximate objectives"—objectives that are feasible enough so as to be achievable by the organization. Rumelt suggests that proximate objectives can help organizations navigate areas of uncertainty.
O’Reilly, Barry. “Optimize to Be Wrong, Not Right.” Barry O’Reilly (blog), April 6, 2017. https://barryoreilly.com/optimize-to-be-wrong-not-right/.