# Scenario planning for the c-suite: beyond best/base/worst
In March 2020, as global supply chains seized and equity markets shed $30 trillion in five weeks, two CFOs made headlines for opposite reasons. At Marriott, Leeny Oberg activated a liquidity playbook within 72 hours, drawing $4.4 billion on revolvers, suspending the dividend, and modeling cash burn under a scenario the team had quietly rehearsed in 2019: "Global health event halts travel for 18+ months." Meanwhile, dozens of mid-cap retailers, whose FP&A teams had built only the classic best/base/worst sensitivity around same-store sales of ±5%, filed Chapter 11 by year-end. The difference wasn't forecasting accuracy. It was scenario *imagination*.
This lesson will dismantle the best/base/worst orthodoxy that still dominates 78% of corporate planning decks (AFP, 2024 survey) and replace it with the Shell-pioneered methodology that has quietly powered the most resilient finance functions of the past five decades.
The three-scenario sensitivity model, typically built by flexing revenue ±10% and holding cost structure roughly constant, survives because it's easy to put in a board deck. It's also analytically useless for two reasons.
First, it confuses *probability distribution* with *strategic alternatives*. When your "worst case" is simply your base case minus 10%, you're not planning for a different world. You're planning for a slightly worse version of the same world. The 2020 pandemic, the 2022 European energy shock (Dutch TTF gas spiked from €20/MWh to €340/MWh in eight months), and the 2023 regional banking crisis (SVB lost $42 billion in deposits in 24 hours) were all "worse than worst case" by the conventional model.
Second, best/base/worst encourages anchoring on the base case. Behavioral finance research from Kahneman's later work and Duke's CFO Survey shows executives unconsciously assign 70-80% probability to the base case even when explicitly told to weight scenarios equally. The result: capital allocation decisions get made as if the base case is reality, and the "worst case" exists only to satisfy the audit committee.
In 1971, a French executive at Royal Dutch Shell named Pierre Wack ran a small team called "Group Planning." Rather than forecasting oil prices, which everyone agreed would stay around $3/barrel, Wack constructed narrative scenarios about what *could* disrupt that consensus: an OPEC cartel asserting pricing power, Arab states using oil as a geopolitical weapon, Western refining capacity becoming a bottleneck.
When the Yom Kippur War hit in October 1973 and oil prices quadrupled to $12/barrel within months, Shell was the only major oil company that had pre-positioned: decentralized refining, flexible supply contracts, and capital reserves earmarked for distressed asset acquisition. Shell moved from seventh-largest oil major to second within a decade.
Wack's core insight: the purpose of scenario planning is not to predict the future. It is to make the organization decision-ready for futures it hasn't yet imagined.
The modern descendant of Wack's method, used by finance teams at Microsoft, Ørsted, and Maersk, abandons linear sensitivity in favor of a two-axis uncertainty matrix. Here's how to build one Monday morning.
Not ten. Not five. Two. From a list of 15-20 driving forces affecting your business, you must identify the two that are simultaneously (a) highest impact on your P&L and (b) most genuinely uncertain. "Interest rates" is rarely one of them, yield curves are forecastable within ranges. Real critical uncertainties for 2026 planning cycles include:
The test: if you can't honestly assign a probability between 30% and 70% to either direction, it's not uncertain enough to anchor a scenario.
Cross your two axes. You now have four quadrants, four genuinely different futures. Name them. (Shell's planners insist on evocative names; "Scenario A" gets ignored, "Fortress Europe" gets debated.)
Each scenario must include: a narrative (one page), a set of leading indicators that would signal you're entering it, and a financial model showing revenue, margin, cash, and capital structure impacts over a 3-year horizon.
This is where most CFOs fail. They build the scenarios, then quietly continue executing the base-case strategy. Wack's rule: for each strategic commitment (M&A, capex, debt issuance), ask: does this decision survive in all four quadrants? If it only works in two, what's the cost of optionality?
Danish energy company Ørsted (formerly DONG Energy) used four-quadrant scenarios in 2012 to model two uncertainties: speed of European renewable subsidies phaseout, and natural gas price volatility. One quadrant, fast subsidy phaseout combined with low gas prices, would have crushed their offshore wind investment thesis. Rather than abandon the strategy, they pre-committed to cost reduction milestones (€100/MWh by 2020) that would make wind competitive *even in that quadrant*. They divested upstream oil and gas for €1.05 billion in 2017, redeployed capital, and by 2019 had hit cost targets five years early. Market cap grew from DKK 98 billion at 2016 IPO to over DKK 400 billion at peak.
The lesson isn't that Ørsted picked the right scenario. It's that they made strategic decisions *robust across scenarios*.
Vérification des acquis
1. According to the AFP 2024 survey cited in the lesson, what percentage of corporate planning decks still rely on the best/base/worst scenario orthodoxy?
2. In March 2020, what specific liquidity action did Marriott's CFO Leeny Oberg take within 72 hours, drawing on a scenario the team had rehearsed in 2019?
3. Behavioral finance research referenced in the lesson (Kahneman and Duke's CFO Survey) shows executives unconsciously assign what probability range to the base case, even when told to weight scenarios equally?
4. Select ALL correct answers describing why the lesson argues best/base/worst is analytically useless.
Sélectionnez toutes les réponses correctes.
5. Select ALL correct answers about the 'worse than worst case' events cited in the lesson as evidence that conventional sensitivity models failed.
Sélectionnez toutes les réponses correctes.
Scenarios that live in a PowerPoint deck reviewed annually are theater. The CFOs who extracted real value from pandemic-era planning embedded scenarios into the monthly operating rhythm. Here's the mechanism.
For each scenario, your team must identify 3-5 leading indicators, observable, measurable signals that you are drifting toward that quadrant. Not lagging KPIs like revenue or churn. *Leading* indicators.
For a SaaS business modeling an "AI commoditization" scenario, leading indicators might include: average sales cycle length (compressing means buyers see less differentiation), gross retention in cohorts under $50K ARRARRAnnual Recurring Revenue (ARR) is the normalized, predictable revenue a subscription business expects to earn from active contracts over a single year.Voir la définition complète →, and competitor pricing in RFPs. For a consumer goods company modeling a "trade war re-escalation" scenario: shipping container spot rates from Shanghai-LA, lead times from Tier 2 suppliers, and customs clearance delays.
Each indicator gets a trigger point, the threshold that activates a pre-defined response. Marriott's 2020 playbook reportedly triggered revolver draws when occupancy fell below 30% for two consecutive weeks. No committee meeting. No debate. The trigger was the decision.
Borrowed from Gary Klein's work and now standard at firms like Bridgewater, the pre-mortem asks: "Assume it's December 2027 and our company has missed its plan by 40%. What happened?"
Run this exercise quarterly with your FP&A team, business unit CFOs, and treasury. The failure modes that surface are almost always uncertainties you've under-weighted. Document them. Promote the top three into formal scenarios.
Microsoft's CFO Amy Hood has publicly described how the company's $80+ billion FY2025 AI infrastructure capexcapexCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.Voir la définition complète → was scenario-tested rather than point-forecasted. The team modeled four worlds across two axes: AI inference demand (high/low) and AI training efficiency gains (fast/slow). In the "high demand, slow efficiency" quadrant, capexcapexCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.Voir la définition complète → was insufficient. In "low demand, fast efficiency," it was overbuilt by ~30%.
Rather than picking one, Microsoft structured contracts with modular commitments, taking firm capacity for 18 months but converting longer-dated commitments into options. When DeepSeek's R1 model in early 2025 triggered concerns that training efficiency had jumped suddenly, the stock dropped, but Microsoft's actual capital exposure was capped by design. Hood's framing in the January 2025 earnings call, "we are building capacity for the demand we see and structuring optionality for the demand we don't", is textbook Wack.
Three uncertainties deserve explicit treatment in any 2026 scenario set:
Pillar Two cash tax volatility. The OECD's 15% global minimum is now in force across 40+ jurisdictions, but the U.S. GILTI/CAMT interaction remains unsettled. Companies with significant IP in low-tax jurisdictions face cash tax swings of 200-400 bps of EBITDAEBITDAEBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) measures a company's operating profitability before financing and accounting decisions, used to compare core performance across firms.Voir la définition complète → depending on resolution paths. Model both convergence and fragmentation scenarios.
CSRD double-materiality enforcement. First mandatory reports were filed in 2025 for FY2024. The question for 2026 is whether assurance providers (and regulators) treat omissions as material weaknesses. Companies should scenario-plan for both light-touch and aggressive enforcement, with implications for sustainability capexcapexCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.Voir la définition complète → timing.
IFRS 16 covenant pressure in a higher-rate world. Lease liabilities sitting on balance sheets at 2020-era discount rates are being remeasured into a 4-5% rate environment, with material impacts on leverage ratios. Stress-test debt covenants against both rate paths and lease renewal timing.
1. Kill the best/base/worst slide in your next board deck. Replace it with a four-quadrant scenario mapmapUsing software to automate repetitive marketing tasks and campaigns, enabling personalisation at scale across channels like email, web, and social.Voir la définition complète → built around two genuine uncertainties. If your audit committee resists, frame it as enhanced risk governance under CSRD Article 19a.
2. Identify three leading indicators per scenario, with pre-committed trigger actions. Document them in a one-page playbook signed off by treasury, FP&A, and the relevant business unit head. The trigger is the decision.
3. Run a quarterly pre-mortem with your top 10 finance leaders. Two-hour session, no slides. Assume catastrophic plan miss; reverse-engineer the failure. Promote the top three failure modes into formal scenarios in the next planning cycle.
4. Audit every committed capital decision over $50 million against all four scenarios. If it only works in one quadrant, you've made a directional bet, not a capital allocation. Either build optionality into the structure (Microsoft model