Robust decision-making accepts that in complex environments the future cannot be forecast with enough precision to justify fine-tuned optimisation. Instead of asking “What is most efficient under our best guess of tomorrow?”, it asks “What holds up acceptably well across many plausible tomorrows?”. The focus shifts from maximising expected performance in a single model of the world to reducing vulnerability and regret across a range of scenarios. This usually means favouring flexibility, redundancy, and options that keep pathways open, even at the cost of slightly lower performance in the most optimistic case.

Practically, working robustly starts with generating multiple, contrasting futures and stress-testing candidate strategies against each. A robust strategy is not the one that wins big in any one scenario, but the one that does not fail catastrophically in any of them, and that can be adapted as signals accumulate. Decision-makers replace the search for the “right” prediction with a search for strategies that remain coherent under different assumptions, data revisions, and shocks. The organisation then monitors simple indicators that suggest which futures are unfolding, and uses them to adjust course, treating plans as provisional scaffolding rather than fixed commitments.