Why Facilitate Scenario Discussions?
A strategic choice can look attractive under the conditions management currently expects and become fragile if those conditions change. Scenario discussion helps decision-makers examine that fragility systematically. Rather than arguing over one preferred forecast, participants consider a small number of plausible, contrasting futures and ask what each would mean for the choices available.
A scenario is not a prediction, nor does it have to be the most likely future. It is a coherent description of conditions that could materially affect a decision. Scenarios are useful when they expose different drivers, reveal assumptions hidden in a proposal, and give leaders a shared basis for comparing choices. They are less useful when they are merely optimistic and pessimistic versions of the same forecast or when discussion becomes an unstructured list of risks.
A well-facilitated discussion has a decision at its center. “What might the market look like?” is too broad on its own. “Should we commit to a fixed-capacity claims operation or retain the ability to use external capacity?” is more useful because it identifies choices and consequences to compare. The discussion should make clear the decision’s time horizon, the outcomes that matter, and any constraints that cannot be traded away.
A Practical Structure for the Discussion
The facilitator’s role is to keep the group moving from plausible conditions to decision implications. The purpose is not to reach consensus about which scenario will occur. It is to determine what the scenarios reveal about the decision, what assumptions drive the comparison, and what remains uncertain.
State the decision, the available options, the time horizon, and the outcomes or constraints that matter. Avoid beginning with a general discussion of risks.
Choose a few plausible scenarios that differ in decision-relevant drivers. State the assumptions clearly and check that each scenario is coherent rather than a collection of unrelated extremes.
For each choice and scenario, identify the relevant benefits, costs, exposures, constraints, and operational implications. Separate known information from judgment and uncertain estimates.
Apply an explicit decision criterion, such as minimum acceptable performance or maximum regret. Ask which assumptions change the ranking and whether a constraint is violated.
Record the assumptions that drive the result, missing evidence, and analyses that could change the decision. Do not disguise uncertainty by assigning unsupported precision.
A scenario comparison can be qualitative or quantitative. If outcomes are quantified, the facilitator should specify what the measure includes and what it leaves out. A table of cumulative net values, for example, should state its time horizon and whether discounting, taxes, or indirect effects are excluded. Consistency matters: comparing one option’s initial cost with another option’s full-horizon cost would not be informative.
Worked Examples: Comparing Scenarios and Choices
Worked Example: Testing a Technology Commitment
A hypothetical insurer is considering three claims-platform choices for a four-year period: a modular upgrade, a full replacement, or deferring the investment. Management defines the comparison measure as cumulative net value in millions of dollars over four years, relative to maintaining the current platform. All stated benefits and costs are already expressed on that basis; for this exercise, no additional discounting is applied. Under a gradual-change scenario, the modular upgrade produces $8 million of benefits and costs $5 million; the full replacement produces $15 million of benefits and costs $9 million; deferral produces neither incremental benefit nor cost. Under a rapid-change scenario, the corresponding benefits are $12 million and $18 million, and the costs are $7 million and $13 million; deferral again has zero incremental benefit and cost. Which choice is robust under a maximin criterion?
Method. Calculate each choice’s net value in each scenario as benefits minus costs, then select the choice with the highest worst-scenario net value. This criterion makes the downside across the stated scenarios explicit; it does not assert that the worst scenario is the most likely.
Gradual change. The modular upgrade has net value \(8-5=3\) million dollars. The full replacement has net value \(15-9=6\) million dollars. Deferral has net value \(0\) million dollars.
Rapid change. The modular upgrade has net value \(12-7=5\) million dollars. The full replacement has net value \(18-13=5\) million dollars. Deferral has net value \(0\) million dollars.
The worst-scenario values are therefore $3 million for the modular upgrade, $5 million for the full replacement, and $0 million for deferral. The maximin choice is the full replacement, with a worst-scenario value of $5 million. It is also best in the gradual-change scenario and tied with the modular upgrade in the rapid-change scenario.
Check. The full replacement’s two scenario values are both $5 million or higher, and its minimum is $5 million. The other choices have minima of $3 million and $0 million, respectively. The maximin comparison is consistent.
Common mistake. Selecting the modular upgrade because it costs less, without comparing benefits, would confuse cost with net value. The calculation uses benefits minus costs for every option and scenario. The conclusion is conditional on the stated outcomes and maximin criterion; different constraints or omitted effects could change it.
Worked Example: Comparing Flexible Capacity With Expansion
A hypothetical insurer handles up to 50,000 claims annually with its current operation. It can either expand internal capacity to 75,000 claims for a fixed incremental cost of $1.8 million per year, or retain current capacity and outsource claims above 50,000 at $300 each. Compare the incremental costs under annual volumes of 48,000, 54,000, and 75,000 claims. Treat the existing operation’s costs as common to both choices and exclude them. Which option is cheaper in each scenario, and at what excess volume do their incremental costs break even?
Method. Expansion costs $1.8 million in every scenario. Outsourcing costs $300 times the number of claims above 50,000, with no outsourcing cost when volume is at or below capacity.
At 48,000 claims, there are no excess claims, so outsourcing costs \(0\times \$300=\$0\). Expansion costs $1,800,000. Outsourcing is cheaper by $1,800,000.
At 54,000 claims, excess volume is \(54,000-50,000=4,000\) claims. Outsourcing costs \(4,000\times \$300=\$1,200,000\). Expansion costs $1,800,000. Outsourcing is cheaper by \(\$1,800,000-\$1,200,000=\$600,000\).
At 75,000 claims, excess volume is \(75,000-50,000=25,000\) claims. Outsourcing costs \(25,000\times \$300=\$7,500,000\). Expansion costs $1,800,000. Expansion is cheaper by \(\$7,500,000-\$1,800,000=\$5,700,000\).
For break-even, let \(n\) be the number of claims above the 50,000-claim capacity. Set the outsourcing cost equal to the expansion cost: \(300n=1,800,000\), so \(n=1,800,000/300=6,000\) excess claims. At 6,000 excess claims, both options cost $1.8 million; above that level expansion is cheaper, and below it outsourcing is cheaper.
Check. The 54,000-claim scenario has 4,000 excess claims, below the 6,000-claim break-even point, and outsourcing is cheaper. The 75,000-claim scenario has 25,000 excess claims, above the threshold, and expansion is cheaper.
Common mistake. Applying the $300 charge to all 54,000 or 75,000 claims would overstate outsourcing cost. It applies only to claims beyond the existing 50,000-claim capacity. The scenario discussion also shows why average demand alone may not settle the choice: the cost comparison changes when volume crosses the threshold.
Worked Example: Using Regret to Compare Strategic Choices
A hypothetical insurer is deciding how to respond to uncertainty about the timing of a new market transition. It considers investing early, using a staged approach, or preserving capital for now. The group estimates net strategic value, in millions of dollars over the decision horizon, as follows: in a rapid-transition scenario, the values are $12 million, $8 million, and $2 million, respectively; in a delayed-transition scenario, they are negative $8 million, $5 million, and $6 million, respectively. Use a minimax-regret criterion to compare the choices.
Method. For each scenario, find the best value available among the choices. Regret is that scenario’s best value minus the value of a particular choice. Then choose the option whose largest regret across scenarios is smallest.
In the rapid-transition scenario, the best value is $12 million. Regrets are therefore: early investment, \(12-12=0\); staged approach, \(12-8=4\); preserve capital, \(12-2=10\) million dollars.
In the delayed-transition scenario, the best value is $6 million. Regrets are: early investment, \(6-(-8)=14\); staged approach, \(6-5=1\); preserve capital, \(6-6=0\) million dollars.
The maximum regret for early investment is \(\max(0,14)=14\) million dollars. For the staged approach it is \(\max(4,1)=4\) million dollars. For preserving capital it is \(\max(10,0)=10\) million dollars. The minimax-regret choice is the staged approach, whose maximum regret is $4 million.
Check. In each scenario, the best choice has regret zero, and no regret is negative. The staged approach has the smallest of the three maximum regrets: $4 million versus $10 million and $14 million.
Common mistake. Choosing the option with the highest average of the two scenario values would introduce an implicit weighting of the scenarios. Minimax regret instead limits the worst missed opportunity relative to the best choice in each scenario. It is one possible decision criterion, not a uniquely correct answer for every organization.
What the Comparison Reveals—and What It Does Not
The examples use numerical criteria to make the logic visible, but the point of facilitation is broader than selecting the largest or smallest number. A scenario discussion should help management distinguish three kinds of result:
- Robust choices: options that meet essential requirements and remain acceptable across the scenarios considered.
- Assumption-sensitive choices: options whose relative attractiveness changes when a specific driver changes, such as claim volume or transition timing.
- Unresolved questions: uncertainties for which better information, operational testing, or additional analysis could materially alter the comparison.
A short derivation explains the maximin calculation used in the first example. Let \(V(a,s)\) denote the value of choice \(a\) under scenario \(s\). The worst-case value for choice \(a\) within the selected scenario set is \(\min_s V(a,s)\). Maximin chooses the option with the largest such minimum:
The inner minimum represents the least favorable outcome for that choice among the scenarios analyzed. The outer maximum selects the choice with the strongest least-favorable outcome. This supports a downside-protection interpretation of maximin, but it does not prove that maximin is appropriate for every decision. A group may instead prioritize expected value when credible probabilities are available, regret, a minimum performance threshold, or a set of non-negotiable constraints. The selected criterion should be stated rather than smuggled into the discussion.
Scenarios also do not automatically justify probability-weighted averages. Unless probabilities are supportable and relevant to the decision, presenting an average can create false precision. Even when probabilities are available, the facilitator should make clear which assumptions produced them and whether the decision is sensitive to them. A scenario’s usefulness comes from testing a choice, not from assigning it a probability by default.
Facilitation Pitfalls and Good Practice
Several habits make a scenario discussion more decision-relevant:
- Keep the scenarios distinct. Give each a clear set of drivers and a plausible internal logic. Varying one label from “low” to “high” while leaving the same assumptions underneath may not test a genuinely different future.
- Separate scenario assumptions from conclusions. Record what participants assumed about conditions before discussing what those conditions imply for each choice.
- Include constraints, not just a summary score. A favorable total value does not make an option acceptable if it breaches a stated risk appetite, capacity, or liquidity constraint.
- Ask what would change the ranking. Identify thresholds and drivers, such as the 6,000-claim break-even point, rather than relying on general statements that an option is “sensitive.”
- Distinguish disagreement from missing evidence. Participants may disagree about objectives, about scenario assumptions, or about the evidence. Those are different issues and should not be collapsed into a single score.
- Document limitations. Note important effects not quantified, ranges that rely on judgment, and scenarios the analysis does not cover.
A useful output is a concise record of the decision question, scenarios, choices, comparison criteria, results, key assumptions, and open analysis questions. It should show both what appears robust and what depends on particular conditions. This makes the discussion actionable without implying that uncertainty has been eliminated.
Check Your Understanding
For each problem, identify the comparison logic and check that the answer follows from the stated assumptions.
- A hypothetical insurer compares choices A, B, and C under two scenarios. Their values, in millions of dollars, are A: $2 million and negative $3 million; B: $0 million and $1 million; C: negative $1 million and $2 million. Under maximin, which choice is selected? (Answer: B; the minimum values are negative $3 million, $0 million, and negative $1 million, respectively.)
- A claims operation has capacity for 40,000 claims. An expansion costs $1.5 million, while outsourcing costs $250 per claim above existing capacity. What excess volume produces equal incremental costs? (Answer: 6,000 excess claims, since $1,500,000 divided by $250 is 6,000.)
- Under two scenarios, values in millions are X: $8 million and negative $2 million; Y: $5 million and $4 million; Z: $1 million and $6 million. Which option minimizes maximum regret? (Answer: Y; maximum regrets are $8 million for X, $3 million for Y, and $7 million for Z.)
- Management assigns no defensible probabilities to three scenarios but asks for a probability-weighted average. What should the facilitator clarify before producing one? (Answer: The scenarios do not by themselves supply probabilities; the group needs supportable probabilities and should state their basis, or use an explicitly non-probabilistic criterion.)
- A choice performs well numerically in every scenario but violates a stated minimum liquidity requirement in one. Is it robust under that requirement? (Answer: No; a breach of a non-negotiable constraint makes the choice unacceptable in that scenario, regardless of its aggregate numerical performance.)