Most strategic planning quietly assumes a single future. A briefing settles on the most likely storyline. A forecast lands as one number. A plan is built for the expected case, and contingencies become an appendix nobody reads. The trouble is that strategic surprise almost never arrives from the future you planned for. It comes from the plausible alternatives no one structured, weighed, or watched for.
This is not a failure of intelligence so much as a failure of form. The single-forecast habit is comfortable -- it produces a clean answer a leader can act on. But the world does not owe us a clean answer, and a confident point estimate can be worse than useless: it concentrates attention on one path while the others develop unobserved.
The single-forecast trap
Three weaknesses recur whenever an organisation collapses an uncertain situation into one prediction.
First, it hides the alternatives. A single storyline cannot represent the branch points where a situation could go one way or another. The fork is exactly where the risk -- and the opportunity -- lives, and it is the first thing a point forecast discards.
Second, it obscures what would change the answer. A number with no visible drivers cannot tell a decision-maker which indicator to watch, which evidence would move the assessment, or what action might steer the outcome. It is an answer with no handles.
Third, it cannot be checked. A forecast delivered as a lone figure, with no record of the reasoning or the evidence behind it, is almost impossible to learn from. Whether it was right or wrong, you cannot say why -- and so the next forecast is no better than the last.
The goal of foresight is not to predict the future with certainty. It is to prepare for more than one of them -- and to know, in advance, what would tell you which is arriving.
Structured foresight: holding many futures at once
The alternative is older than any software, and it is the tradition serious foresight shops have used for decades: structure the problem before you try to answer it. Done well, it rests on a few complementary disciplines, each compensating for the others' blind spots.
Map the space of futures. Rather than guessing the single most likely outcome, lay out the dimensions of the problem and the states each can take, then systematically work through the combinations -- ruling out the ones that cannot coexist. What remains is the set of internally coherent futures worth taking seriously. Nothing credible is quietly dropped; nothing incoherent is allowed to distract.
Reason about likelihood as evidence arrives. Not every future is equally probable, and the balance shifts as new information comes in. Probabilistic reasoning weighs the alternatives and updates them transparently, with the drivers behind each assessment on view rather than buried in a model's internals.
Model the dynamics. Situations are not snapshots. Feedback loops amplify small moves; delays mean causes and effects are separated in time; thresholds turn a stable picture volatile in a single step. System-dynamics modelling captures how a situation actually evolves, so foresight reflects motion, not a frozen frame.
Ground everything in evidence. A scenario is only as good as the facts beneath it. An evidence-gated intelligence pipeline keeps each assessment tied to sourced, relevance-checked material -- and keeps a link from the assessment back to the evidence that supports it, so the reasoning can be retraced later.
Where AI fits -- and where it does not
It is tempting, in 2026, to hand the whole problem to a large language model and ask it for the future. The results are fluent, fast, and frequently wrong in the most dangerous way: they are plausible. A model left to narrate the future on its own produces confident stories that smooth over exactly the structural surprises foresight exists to catch.
The discipline matters more than the model. AI is genuinely transformative inside a structured method -- accelerating the construction of scenarios, the integration of evidence, and the bookkeeping of probabilities at a speed no analyst could match by hand. But it accelerates the method; it does not replace it. Used to run structured foresight faster, AI is a force multiplier. Used as a substitute for structure, it is a confident guess in a nicer wrapper.
Foresight you can audit -- and decisions that stay human
Two principles separate decision support that earns trust from a black box that merely demands it.
The first is transparency you can audit. When every assessment carries its drivers and its evidence, an analyst can see why the picture changed, not just that it did. And when assessments are recorded over time, foresight becomes something rare in this field: a practice that can be measured against what actually happened, and improved on that basis, rather than asserting a track record it never shows.
The second is that the decision stays human. Structured foresight is decision support, not decision automation. Its job is to widen the field of view, sharpen the questions, and surface the options that hold up across the most futures -- then hand a clearer choice to the person accountable for making it. The machine structures; the human decides.
From situation to decision
This is the conviction behind BlackDecision, our decision-intelligence platform: that the right response to an uncertain world is not a better single guess, but a disciplined way to hold many futures at once -- reasoned about probabilistically, modelled as they evolve, grounded in evidence, and kept under the decision-maker's command. Anticipation, made structured. The future will still surprise us. It should surprise us a great deal less often.