Scenario planning is a structured way of testing decisions against multiple plausible futures. It becomes useful only when each scenario forces the Decider to name the assumptions that would have to hold for the decision to succeed. Without that discipline, scenario planning is speculation with a budget.
I do not object to the discipline. I object to the respectable fiction that passes for it in many organisations: the handsome workshop, the polished deck, and no decision anybody can defend. In 1972 and 1973 Shell used scenarios well. The point was not prophecy. It was to keep choices alive and to test what management was assuming before the oil shock arrived.
What scenario planning is (and is not)
At its best, scenario planning is a disciplined way to test a tentative decision against several plausible futures. Shell did not prosper because it guessed October 1973 like a fortune teller. It prospered because managers had already considered how their choices would fare if oil politics turned ugly, and because they had not married themselves to one comforting story about the future. Shell matters as proof that good Deciders keep options open.
The copied version forgot the hard part. It kept the scenarios and dropped the assumption work. That is why so much scenario based planning becomes a creative writing exercise with a catering budget. People spend a day naming futures, give each one a memorable label, and leave without asking what would have to be true for their decision to succeed in any of them.
Once that question is missing, the exercise stops serving Purpose. It becomes theatre for the strategy team, reassurance for the board, or billable hours for somebody with a slide template. I have seen all three. None of them help the Decider who must commit money, reputation, or operating capacity on Monday morning.
Why most scenario planning fails
The common scenario planning process fails before it starts because it begins with futures instead of Purpose. A decision has to serve something. If we have not said what outcome we are trying to achieve, over what period, and in what Context, then the whole workshop floats free of the real decision. That defect is common in the glossy methodology sold by consultants. It gives you atmospherics before it gives you a question.
The second defect is quieter and more dangerous. Nobody names the assumptions. A future is only a story until we ask, plainly, what would have to hold for this option to work here. That is why I keep returning to the same trigger question. It does more work than a wall full of sticky notes.
The third defect is that every variable gets treated as equally interesting. The usual critical-uncertainties matrix flatters the workshop because everyone gets to contribute a card, then the cards get sorted into a neat square. The decision does not care about neat squares. It cares about which assumptions are load-bearing and weak.
I have taken that matrix apart properly in a separate piece on critical uncertainties scenario planning, including what I rank instead of drivers.
The fourth defect comes after the applause. Nothing is watched. No premise is named, no threshold is set, no response is agreed. That is the difference between real decision work and fiction dressed as governance. The exercise ends when it should be turning into monitoring. Forecasting versus scenarios is almost the wrong comparison here. Both can fail for the same reason: they hand the Decider a story about the future instead of a judgement about the assumptions under the decision.
Where scenario planning works and where it fails
Good scenario work does not try to sound clever. It tries to be usable. The pattern is the same in the better cases I covered in decision making under deep uncertainty: several plausible futures, explicit assumptions, and a way to revise when the premises move.
I have written a longer piece on scenario planning examples that actually changed decisions, grading each case against the discipline above. The pattern holds: the useful ones left an owned trigger behind.
Colorado River. The Bureau of Reclamation tested 8,400 hydrologic scenarios because the river serves 40 million people and no single forecast was honest enough. The work mattered because it looked for strategies that performed acceptably across multiple futures, not because 8,400 is an impressive number.
Netherlands Delta Programme. The Dutch did not build one majestic long-range plan and hope for the best. They tied adaptation strategies to monitoring triggers, which is exactly how a decision stays alive after the planning documents are filed.
Louisiana Coastal Master Plan. Louisiana tested 124 projects across three environmental scenarios over a 50-year problem and built revision into the plan from the start. That is planning without prophecy.
Bad cases fail for the opposite reason. The future was not the problem. The buried premise was.
TEPCO and Fukushima. Internal modelling in 2008 showed tsunami run-up above 15 metres against a 5.7 metre seawall, but the design assumption stayed buried until the sea answered it. The organisation had information and still would not force the premise into the open.
Long-Term Capital Management. The fund's models assumed continuous market liquidity and stable correlation patterns across asset classes. When Russia defaulted in 1998, both assumptions failed simultaneously. The models were sophisticated. The premises beneath them were invisible. Nobody had asked whether those two conditions could break together.
NASA Columbia. Foam had struck the orbiter's heat shield on previous flights without catastrophe, so the assumption that foam strikes were survivable became institutional fact. During the 2003 mission, engineers raised concerns and were overruled by a process that had long since stopped questioning the premise. The shuttle broke apart on re-entry. Process without assumption testing is scenery.
The scenario planning framework that actually works
If you want a scenario planning framework that does more than decorate a board paper, it has to sit inside the Universal Decision-Making Method. That is the discipline Roger Estall and I kept arriving at, and it is still the only version I trust. Most models stop at the picture. The method keeps going until somebody can decide.
1. Frame the decision. Start with Purpose. What are we trying to achieve, over what duration, and in what Context? If that is vague, the scenarios will be vague as well, because they are not anchored to a real decision.
2. Develop options. Use scenarios to test tentative decisions, not to admire the future. If the option set does not change after the scenarios are discussed, the exercise has produced very little.
3. Recognise assumptions. Ask the blunt question: "Okay! What are the assumptions we are making here?" Then write them down. That move turns the exercise from speculation into decision work, because we can now see what each option rests on.
4. Sufficient certainty. Judge each assumption by two things: how much influence it has on the desired outcome, and how much confidence we have that it will hold. Now we know which assumptions are critical, which are merely interesting, and whether we have enough certainty to proceed. I do not ask for maximum certainty here. I ask for enough certainty to decide well.
5. Design monitoring. Name the assumptions worth watching. Then say what movement would matter and what we will do if it occurs. The method does not end with a workshop. It ends with a live decision that knows what to watch.
Most scenario planning tools have the same weakness as the market for risk registers. They sell apparatus. Give people a quadrant, a scoring screen, or a dashboard and they feel industrious. They have still not named the assumptions beneath the decision. A useful tool should force the Decider to record the tentative decision, record the assumptions attached to it, judge their influence and confidence, and then record what will be watched afterward. If the tool cannot do that, it is a drawing aid. I am not hostile to software. I am hostile to software that helps people avoid the only hard question.
From scenarios to decisions, and what to watch after
The bridge from scenarios to decisions is where most organisations lose their nerve. They keep working the futures because it feels safer than choosing. Good strategic scenario planning stops earlier. It asks whether we have sufficient certainty, not whether we have wrung every last ambiguity out of the future.
I have seen organisations pursue maximum certainty on a single visible risk and create a different problem entirely: supply shortages, capability gaps, delays that erode the outcome the decision was supposed to protect. That is why the Universal Decision-Making Method asks for sufficient certainty, not maximum certainty. The standard is enough to decide well, not enough to eliminate every worry from the room.
When we do not have sufficient certainty, there are only three honest responses. We can get more information. We can modify the decision with secondary elements such as staged commitments, price terms, or fallback capacity. Or we can decide that the option should not go ahead. None of that is glamorous. All of it is real.
Once the assumptions are named, the decision usually gets smaller and plainer. We stop admiring the future and start judging the premises.
The monitoring that follows should be equally plain. A named premise, a trigger, and a response. That is far better than the generic recovery scenarios I criticised in disruption, where organisations keep a folder of dramatic events and no live rule for what to watch this month.
The real danger is often slower than the workshop imagined. The boiling frog problem is not cinematic. A cost assumption drifts. A policy assumption hardens. A customer assumption stops holding. Because nobody named the premise, nobody sees the failure until the outcome has already moved. The discipline earns its keep only when it leaves behind a short list of live assumptions and the habit of revisiting them before the water boils.
I have been making this argument for years, and I make it again here. Scenarios are respectable. Unnamed assumptions are not. When the exercise helps us see what must be true, decide whether that is sufficiently certain, and watch the premises afterward, it is worth doing. When it does not, it is respectable fiction.
Grant Purdy is the co-author, with Roger Estall, of Deciding (2020), and the architect of the Universal Decision-Making Method.
If you have a decision you are working through, the Walk can help.
Start a Walk