Scenario Planning: Preparing for Futures That Refuse to Pick One Path

Scenario planning is a strategic-planning discipline for deep uncertainty: instead of forecasting one future and betting on it, you construct a small set of structurally different, internally consistent futures and test strategy against all of them. The lineage runs from Herman Kahn's scenario work at RAND in the 1950s–60s through Pierre Wack's team at Royal Dutch Shell, whose scenario practice before the 1973 oil shock became the canonical business case, described in Wack's two 1985 Harvard Business Review articles ('Scenarios: Uncharted Waters Ahead' and 'Scenarios: Shooting the Rapids'). The classic process: frame the focal decision and horizon; identify driving forces (a PESTLE scan feeds this); separate predetermined elements from critical uncertainties; pick the two most important, most uncertain, independent drivers as axes of a 2×2 matrix; develop the four resulting futures into named, narrated scenarios; test the strategy against each; and derive robust moves, hedges, options, and early-warning indicators. Known failure modes: scenarios as entertainment with no decision consequence, one 'official future' plus decorative variants, probability assignments that collapse the exercise back into forecasting, and indicator lists nobody monitors. On an argument tree, each scenario becomes a branch of structured claims: 'strategy X survives scenario A' is a claim with supporting and attacking arguments, robustness emerges as visible cross-scenario support, and when an early indicator fires, the affected branches update. In decision-quality terms, scenario planning feeds the frame and alternatives elements; the argument tree supplies the sound reasoning that turns four stories into one defensible strategy.

All decision frameworks
Framework guide · strategy under uncertainty

Scenario Planning

In 1973 the oil price did the impossible, and one company had already rehearsed it. The discipline Pierre Wack built at Shell — and how to run it so the scenarios change decisions, not just slides.

TL;DR

Scenario planning replaces one forecast with a small set of structurally different futures, then tests strategy against all of them:

  • Separate the predetermined from the uncertain — Wack's core move: some things are already in the pipeline; the exercise is about the rest
  • Two critical uncertainties → a 2×2 → four named futures, each internally consistent, none of them 'the official one'
  • Strategy is tested against all four: robust moves survive everywhere; hedges and options cover the rest; indicators say which world is arriving
  • On an argument tree, 'we survive scenario A' is a claim to attack — robustness becomes visible support, not asserted comfort

What scenario planning is — and the story that made it famous

Scenario planning starts from an admission most planning avoids: for some decisions, the future genuinely branches, and no amount of analysis will collapse it to one line. Rather than betting on a single forecast, you build a handful of structurally different, internally consistent futures and ask of your strategy: how does it fare in each?

The method's intellectual roots are in Herman Kahn's scenario work at RAND in the 1950s–60s — thinking the unthinkable about nuclear strategy by narrating it. Its business canonization came at Royal Dutch Shell, where Pierre Wack's planning group built scenarios in the early 1970s that included a future of producer-driven oil-price shocks. When the 1973 embargo hit, Shell's managers had already rehearsed the world they woke up in. Wack told the story — and the method's real lesson — in two 1985 Harvard Business Review pieces, "Scenarios: Uncharted Waters Ahead" and "Scenarios: Shooting the Rapids".

Wack's lesson is routinely missed by imitators: the scenarios' purpose is not prediction but re-perception — changing the mental models of the managers who must decide. A scenario deck that doesn't alter what anyone decides is, in Wack's framing, water on a stone. That standard — decision consequence — runs through everything below. Context in the toolbox: decision-making models; scenario planning's raw material comes from a PESTLE scan.

When to use it — and when not to

Scenario planning earns its considerable cost when:

  • The decision is long-lived and the environment genuinely branches. Infrastructure, platform bets, market entries with decade horizons — where being wrong about the world costs more than the exercise.
  • Key uncertainties are identifiable but not probabilizable. You can name the drivers (regulation tightens or not; a technology matures or stalls) without honestly being able to weight them.
  • The leadership's mental model is the risk. When everyone plans inside one 'official future', scenarios are the instrument that makes alternatives thinkable — their original Shell function.

And the failure modes — scenario planning fails more often than it works, usually one of these ways:

  • Scenarios as theater. Four vivid futures, a stimulating workshop, zero changed decisions. If no strategy, hedge, or trigger changed, the exercise failed by Wack's own standard.
  • The official future plus decoration. One scenario is the real plan; the others exist to make it look considered. The tell: capital allocation matches exactly one future.
  • Probability creep. Assign likelihoods and people optimize for the 60% world — collapsing the method back into the single forecast it was built to escape.
  • Indicators nobody watches. Early-warning signposts defined in the workshop and never monitored. A scenario system without a watch rota is a document, not a capability.

Step by step, with a worked example

Illustrative scenario: an invented European logistics company setting its five-year fleet strategy. The procedure:

  1. 1Frame the focal decision and horizon. "What fleet and charging infrastructure do we commit to over five years?" Scenarios without a focal decision drift into futurology.
  2. 2Identify driving forces. A PESTLE-style scan surfaces them: emissions regulation, battery cost curves, charging build-out, freight demand, labor supply, fuel prices.
  3. 3Split predetermined from uncertain. Wack's crucial move. Already in the pipeline: announced low-emission-zone rules, the age profile of the current fleet. Genuinely uncertain: pace of charging infrastructure; freight-demand structure (e-commerce vs industrial mix).
  4. 4Pick two axes, build the 2×2. Axis 1: charging infrastructure — fast build-out vs stalled. Axis 2: freight demand — fragmented/urban vs consolidated/long-haul. Four futures, each named and narrated until internally consistent: Wired & Local, Wired & Heavy, Stalled & Local, Stalled & Heavy.
  5. 5Test the strategy in each world. The all-electric-now plan thrives in the Wired futures and strands capital in the Stalled ones. The all-diesel plan inverts. A staged mixed fleet with conversion-ready depots survives all four — worse than the best bet in its best world, catastrophic in none.
  6. 6Derive moves, hedges, options, indicators. Robust move: depot conversion-readiness (cheap now, valuable everywhere). Option: contracted capacity with an EV supplier, exercised only in Wired worlds. Indicators with owners: quarterly charging-corridor build rate; e-commerce share of contract mix. Each indicator names who watches it and which decision it triggers.

Scenarios as an argument tree

In decision-quality terms, scenario planning feeds the frame (which world are we deciding in? — answer: several) and alternatives (strategies you'd never generate inside the official future). What the four narratives lack is a place where 'our strategy survives scenario A' gets examined rather than asserted. On an argument tree:

Strategy → root claim

"Commit to the staged mixed fleet." The recommendation stands at the root; the scenarios organize the case around it.

Each scenario → a branch of claims

"The staged fleet remains viable in Stalled & Heavy" is a claim, with supporting arguments (diesel capacity retained) and attacks (conversion capex partly stranded) — argued, not narrated.

Robustness → visible cross-branch support

A strategy that genuinely survives all four worlds shows four supported branches. One that secretly banks on the official future shows it — one strong branch, three thin ones.

Indicators fire → branches update

When the charging build-rate indicator crosses its threshold, the affected scenario branches take new evidence and the root's support shifts — the watch rota has somewhere to land its findings.

The one-sentence version

Scenarios supply frame and alternatives; the argument tree supplies the sound reasoning that turns four stories into one strategy you can defend in all of them. See decision quality.

Scenario planning vs the alternatives

If your question is…Reach forWhy not scenarios
The uncertainties are estimable — approval odds, demand rangesDecision tree analysisProbabilizable uncertainty deserves arithmetic, not narratives
What macro forces are even out there?PESTLE analysis (feeds the scenario axes)PESTLE lists drivers; scenarios combine them into worlds
Which risks do we manage continuously?Enterprise risk managementERM runs a register; scenarios stress whole strategies
Keeping the door open has value in itselfReal options thinkingOptions price the flexibility scenarios reveal

Frequently Asked Questions

What is scenario planning?

A strategic-planning discipline for deep uncertainty: instead of forecasting one future, you construct a small set of structurally different, internally consistent futures and test your strategy against all of them. The output is not a prediction but a strategy portfolio — robust moves that work everywhere, hedges and options for specific worlds, and early-warning indicators that tell you which future is arriving. The method descends from Herman Kahn's work at RAND and was canonized in business by Pierre Wack's scenario team at Shell before the 1973 oil shock.

What did Shell actually do with scenario planning?

Pierre Wack's planning group at Royal Dutch Shell built scenarios in the early 1970s that included a then-heretical future: oil producers restricting supply and driving prices up sharply. When the 1973 embargo made that world real, Shell's managers had already rehearsed responses to it. Wack's own account, in two 1985 Harvard Business Review articles, stresses the real mechanism: the scenarios' value wasn't predicting the shock but changing managers' mental models beforehand — 're-perceiving' the world so the response didn't have to wait for the disbelief to fade.

How do you build a scenario matrix (2×2)?

First separate predetermined elements — things already in the pipeline, which belong in every scenario — from genuine uncertainties. From the uncertainties, choose the two that are most important to your focal decision, most uncertain, and reasonably independent of each other. Their extremes form the axes of a 2×2, giving four candidate futures; develop each into a named, internally consistent narrative, discarding combinations that can't cohere. Four is a convention, not a law — but fewer forces sharper thinking than a fan of ten mini-forecasts.

Should you assign probabilities to scenarios?

The classical answer is no, and the reason is behavioral: assign likelihoods and decision-makers immediately optimize for the 60% world, collapsing the method back into the single-forecast planning it was designed to escape. Scenarios earn their keep precisely on uncertainties too deep to weight honestly. If your uncertainties are estimable, use a decision tree instead — that's the right tool for probabilizable branching. What replaces probability in scenario work is monitoring: early-warning indicators, each with an owner and a trigger, that reveal which world is actually arriving.

Why do most scenario planning efforts fail?

Because they stop at the stories. The recurring failure modes: scenarios as workshop entertainment with no decision consequence; an 'official future' the capital plan actually follows, with the other scenarios as decoration; probability creep collapsing the set back into a forecast; and indicator lists defined once and never watched. The fix is Wack's own standard — decision consequence: every scenario exercise should end with named robust moves, world-specific hedges or options, and monitored indicators that trigger revisits. If nothing changed in what anyone decides or watches, the exercise failed.

How does scenario planning work on an argument tree?

The strategy recommendation sits at the root, and each scenario becomes a branch of claims about it: 'the strategy remains viable in this world' is a claim with supporting and attacking arguments, rated by participants. Robustness stops being asserted and becomes visible — a genuinely robust strategy shows support across all branches, while a disguised official-future bet shows one strong branch and three thin ones. When a monitored indicator fires, the affected branches take new evidence and the root's support shifts, keeping the scenario system alive between planning cycles.

Related frameworks

Test your strategy in four worlds at once

Scenarios as branches, survival claims as arguments under attack, and indicators that update the case when the future starts choosing.

Start Free — No Credit Card

Free forever for individuals