Consensus is treated, in most organizations, as an unambiguous good: more agreement is better than less, and time spent building it is time well spent. That's true often enough to become a habit, and like most habits, it stops getting questioned exactly when it should be.
Consensus is expensive in a specific way that's easy to underweight: it's expensive in time, and time has a cost that compounds the longer a decision sits open. A decision that takes six extra weeks to build full agreement around isn't free during those six weeks. Competitors move, conditions change, and the people waiting on the decision either stall or, worse, start making their own smaller decisions to fill the gap, which then have to be unwound later.
The decisions worth that cost share a common feature: durability matters as much as correctness. A decision that a large group of people had real input into and genuinely committed to will survive implementation pressure that a technically superior decision, imposed quickly, often won't. Reorganizations, major capital commitments, and anything that depends on sustained behavior change across a large group tend to fall into this category.
The decisions that don't deserve the same investment are the ones where speed itself is the primary source of advantage, or where the decision is reversible enough that a fast, slightly-wrong call beats a slow, fully-agreed one. Knowing which category a given decision falls into, before defaulting to the consensus-building process an organization always uses, is a judgment call in its own right, and one that's worth making deliberately instead of by habit.