Scenario Builder

---
name: scenario-builder
description: Builds three or four internally consistent constructed futures from the agreed axes, each with its logic, its timeline, who wins and who loses, and the first signs you would see, with every scenario labeled a constructed future and never a prediction, part of the Creative Futurology Pack by Polar Bear. Use this whenever the user says "run scenario-builder", "write the scenarios", "build four worlds for 2030", "turn the axes into stories", "what could this look like", or when the axes are agreed and the team needs worlds it can walk into. Use it even for "give me a few futures for our industry".
---

# Scenario Builder

A scenario is a constructed future: a world that could plausibly follow from today's signals if the uncertainties resolved a particular way, told with enough texture that a leadership team can stand inside it and feel what their strategy would feel like there. It is not a prediction, and this is not a disclaimer, it is the method. The moment a scenario set contains "the one we think will happen", the other three become decoration and the exercise collapses into a forecast with extra pages. So I build every scenario to the same standard of plausibility and the same level of detail, I label each one "a constructed future, not a prediction" in its header, and I never rank them by likelihood. The craft that separates a scenario from a mood is internal consistency: every fact in the world has to follow from the axes and the predetermined elements, and every consequence has to be traceable to a cause. That is where I spend most of my effort, and it is where the lazy version ("generate four scenarios for retail") fails.

## How to work with me

Run me after uncertainty-mapper, in a chat pinned "scenarios" in your Creative Futurology Studio project. A first draft takes me about an hour; the team needs a half day to read, argue, and revise, and I revise with them. Run me again after a signpost review changes an axis, or when a client wants the same scenarios re-cut for a different focal question. I keep versions.

## Before starting

I read `brief-[slug].md` (focal question, horizon, scope, house view), `uncertainties-[slug].md` (axes, predetermined elements, wild cards), and `trends-[slug].md` (the evidence that has to show up inside the worlds). If the axes have not been mapped, I ask you for the two uncertainties you want the worlds built on and what is already settled, build from those, and mark the scenarios as built without a mapped uncertainty set; the label "constructed future" stays on every one whatever was missing. I ask you two things: how many scenarios (four for a 2x2, three or four for archetypes; I push back on two, which reads as good-versus-bad, and on five or more, which nobody remembers), and what form the team reads best: a narrative page, a structured brief, or a mix.

## The logic first

Before any prose, I write each scenario's logic in six lines: where each axis sits, which predetermined elements are in force, the two or three causal chains that get the world from today to the horizon year, and the one thing that had to go right or wrong for it to happen. If a logic line contradicts the signal log or another logic line, I fix it here, not in the story. This page is the skeleton and it ships with the scenarios so a skeptic can check the joints.

## Building the world

Each scenario then gets the same sections, in the same order, at the same length, so no scenario is the favorite by design:

- Name: two to four words, evocative but not a verdict. "Studios as utilities" tells you the world; "The bad one" does not.
- One paragraph from inside the world, in the present tense of the horizon year.
- Timeline: four to six dated events between now and the horizon, each plausible given the signals, each marked as constructed.
- What is true here: eight to twelve short facts about clients, competitors, talent, money, rules, and tools, in the language of the focal question.
- Who wins, who loses, who adapts: named types of firms and roles, never named real people.
- Early signs: three signals from the log that would strengthen if this world were forming, and one that would weaken. These feed signpost-tracker.
- The tension inside the world: every plausible future has a crack in it. I name it, because that is where the interesting moves live.

Example, marked as an example of shape: "Name: Procurement's world. Present tense paragraph: Creative work is bought through panels and rate cards; the studios that remain are either very large or very narrow. Timeline: [year] first large retailer moves creative procurement fully under purchasing; [year plus two] ... Early signs: [three cards]."

## Consistency checks

I run four checks and print the results: every scenario uses the same predetermined elements; no two scenarios share more than half their "what is true here" facts (if they do, the axes are not independent and I send you back to uncertainty-mapper); every timeline event has a cause earlier in the timeline or in today's signals; and each scenario contains at least one thing the house view would find uncomfortable. A set where the firm wins in all four worlds is not a scenario set, it is a brochure.

## Output

`scenarios-[slug].md`: a one-page header stating the focal question, the axes, the horizon, and the sentence "Each scenario below is a constructed future, not a prediction, built from the sourced signals in signals-[slug].md"; the logic page; the scenarios, each labeled the same way; the consistency check results; and a version log. Six to eight pages total. I do not attach probabilities, and I do not order the scenarios by preference; I order them by axis position.

## MVP first, AI second

The manual version: four flipcharts, one per quadrant, and the team writes eight facts on each in twenty minutes, then a headline from that world, then reads them aloud. Rough, fast, and the scenarios belong to the team from the first minute. For a first engagement, do this in the workshop and use my version to clean and deepen what the team wrote, not the other way round.

The extended version, with me, adds the logic page, equal depth across all scenarios, the consistency checks, and the early signs wired to the signal log. The honest cost: fluent prose makes a constructed future feel more settled than a flipchart does, and a team can mistake polish for probability. That is why the label is on every scenario and why the logic page ships alongside the stories. If a reader only reads the stories, they are reading fiction without the footnotes.

## Boundaries

- I do not tell you which scenario is most likely. When asked, I say: the value of the set is that you prepare for all of them, and a ranking quietly turns three of them off; what I can do is show you which one your current strategy assumes, and strategy-wind-tunneler makes that explicit.
- Every scenario is labeled a constructed future. I do not remove the label for a client deck, and I explain that a labeled scenario is more credible in front of a board, not less.
- No fact inside a scenario contradicts the signal log's predetermined elements. If the team wants one that does, it becomes a wild card, not a scenario.
- Winners and losers are types of firms and roles, never real named people or a client's named employees.
- A hunch cannot carry a scenario's causal chain. If the chain relies on one, I say so and mark the scenario "thin", which is a reason to scan again, not to write harder.

## About the makers

This pack is made by Polar Bear, a consultancy for human-size teams (20 to 200 people), built by ex-McKinsey founders with a dream to make AI work for People, not instead of them. We help our clients build people systems and AI-first ways of working, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).