Scenario Sets

A single analysis often benefits from more than one way of framing the future. Different stakeholders may want to explore different axes of uncertainty, or an analyst may want to iterate on a set of scenarios without discarding earlier work. Scenario sets let you maintain multiple independent collections of scenarios within the same analysis, each with its own constraints, likelihoods, and forecaster input.

What Is a Scenario Set?

A scenario set is a named group of scenarios that belong to a single analysis. Every analysis starts with one scenario set - the primary set - and you can create additional sets at any time. Each set is self-contained: it has its own title, its own list of scenarios, and its own constraint settings.

Additional scenario sets let you:

  • Explore alternative framings of the same issue using different drivers or uncertainty axes
  • Preserve a baseline set of scenarios while experimenting with variations
  • Maintain separate sets for different audiences or time horizons
  • Compare how different scenario structures lead to different likelihood assessments

The Primary Set

Every analysis has exactly one primary scenario set. The primary set is the default - it is the set that appears when you first navigate to an analysis's scenarios, and it is the set referenced by generated products such as action memos and one-pagers. When an analysis is created, Hinsley automatically creates a primary set for it.

You can change which set is primary at any time. Promoting a different set to primary automatically demotes the previous one. The primary set cannot be deleted; if you want to remove it, first promote another set to primary.

Scenario Settings

Each scenario set carries two constraint flags that describe the logical relationship between its scenarios:

  • Mutually exclusive - The scenarios do not overlap. Only one of them can occur. This is the classic “which future will we end up in?” framing. On its own, it means the likelihoods total 100% or less, with the remainder left for outcomes the set does not list.
  • Exhaustive - The scenarios cover all reasonable possible outcomes. Combined with mutual exclusivity, this makes the set a complete partition of the future possibility space, and its likelihoods total exactly 100%.

New scenario sets start with both flags enabled. This includes the primary set that Hinsley creates with an analysis.

When Hinsley generates AI likelihoods for a scenario set, it factors the constraint settings into its reasoning. A set that is both mutually exclusive and exhaustive gets likelihoods that total exactly 100%. A set without either flag gets likelihoods that are independent of one another.

You can change the constraints for a set at any time from the scenario settings panel. If the set already has scenarios, Hinsley will offer to regenerate likelihoods to reflect the updated constraints.

Likelihoods in a Mutually Exclusive and Exhaustive Set

In a set that is both mutually exclusive and exhaustive, Hinsley treats the likelihoods as one allocation across the whole set. Each published allocation gives every scenario a likelihood. The AI likelihoods, the aggregate of all estimates, and forecaster estimates on an account with granular likelihoods enabled each total exactly 100%. A reader never sees an allocation that is only partly updated.

For AI likelihoods, Hinsley accepts a generated allocation whose total is within two percentage points of 100%. It then rounds the values to whole numbers that total exactly 100%. If a generated allocation is incomplete or further from 100%, Hinsley rejects it, and the previous AI likelihoods stay in place.

Forecasters submit estimates for every scenario in the set together. Editing the estimate for one scenario opens the form for the whole set. What the form accepts depends on the account's Granular Scenario Likelihoods setting:

  • Granular likelihoods enabled - Each estimate is a whole-number probability, and the estimates must total exactly 100%. The form shows a running total as you type. The Auto-sum to 100% link scales the entered values proportionally so that they total 100%.
  • Granular likelihoods disabled - Each estimate is a likelihood rating. Hinsley uses the center of each rating's range when it builds the aggregate, which it scales to total 100%.

If the scenarios in the set change while a forecaster is filling in the form, Hinsley rejects the submission. The forecaster reloads the page and submits the estimates again.

Cloning a Scenario Set

Rather than building a new set from scratch, you can clone an existing one. Cloning copies all scenarios and constraint settings into a new set. The clone is independent of the original - editing scenarios in one set does not affect the other.

Cloning applies when you want to:

  • Branch off a working set to test a different framing without changing the original
  • Create a snapshot of your current scenarios before a major revision
  • Give a collaborator their own copy of a set to modify independently

Exporting a Briefing

A briefing is an editable PowerPoint deck that compares a scenario set at the start and end of a period. It shows how the scenario likelihoods moved, and how the forecasting questions on each driver of the analysis’s primary decomposition moved. To export one, open the settings menu on the Scenarios page and select Export briefing.

Export briefing is disabled until every scenario in the set has an AI likelihood. If you add a scenario after the last AI likelihood update, the item stays disabled until Hinsley generates AI likelihoods that include the new scenario.

Choosing the Period

The export dialog asks which period of changes the briefing shows:

  • Since last briefing - Starts where the most recent completed briefing for this scenario set ended. This option appears only after a briefing for the set completes.
  • Last 4 weeks
  • Last 13 weeks
  • Since launch - Starts at the first AI likelihoods for the current scenarios.

The dialog selects Last 4 weeks by default. If the last briefing ended 7 or more days ago, it selects Since last briefing instead.

A period cannot start before the likelihood history of the current scenarios. If the chosen period starts earlier than the first AI likelihoods, or earlier than the creation of the current scenarios, the briefing starts at that later date. A footnote in the deck states the date.

What the Briefing Contains

  • Cover - The analysis title, the date of the briefing, the start of the period, the number of scenarios, and the name of the user who exported it.
  • Bottom line - A headline and the current likelihood of each scenario compared with its likelihood at the start of the period.
  • Trajectory - A chart of the weekly scenario likelihoods, with markers for key events and an explanation of why the outlook changed. Key events are documents published inside the period.
  • Drivers - The movement of each driver's forecasting questions. A card describes each driver that moved. A driver that changed little gets a table row with its name, its current state, and a “Little change” label.
  • What to watch - Forecasting questions and indicators to follow next. The deck omits this slide when there is nothing to list.
  • Driver detail - One appendix slide for each driver, with its latest assessment and the start and current values of its forecasting questions.
  • Key sources - The sources that the research update summaries cited most often in the period. Titles link to the original source.
  • Method - How the likelihoods are produced, and counts of the questions, forecasts, sources, and research update summaries behind the briefing.

A move counts as a change when it is more than 5 percentage points or when it crosses a likelihood band. The briefing shows smaller moves as steady.

Every figure in the deck comes from the likelihood and forecast history that Hinsley stores. Hinsley AI writes the headline, titles, and explanations, and the figures in that text are filled in from the same stored data. If Hinsley cannot produce text that matches the data, the export fails.

The briefing is written in the language of the analysis creator, because the scenarios, drivers, and summaries are stored in that language.

Downloading and Editing

Hinsley generates the briefing in the background and opens the Your Exports page. Download the .pptx file from that page when its status is Completed.

All text, tables, and shapes in the deck are editable. The trajectory chart is a native PowerPoint chart, and its data is in a workbook embedded in the file, so you can edit the chart data in PowerPoint.

Scenario Generation & Quality Validation

When Hinsley generates a new set of scenarios, it does not stop at the initial draft. It automatically runs a quality review to check whether the generated scenarios satisfy MECE principles - that they are sufficiently distinct from one another (mutually exclusive) and together cover the reasonable possibility space (collectively exhaustive).

The generation workflow proceeds in three stages:

  1. Generate - Hinsley produces an initial set of scenarios based on the analysis topic, constraint settings, and source material.
  2. Validate - An automated review checks for pairwise overlaps between scenarios, coverage gaps (important futures that no scenario captures), and abstraction or time-horizon mismatches.
  3. Heal - If issues are found, Hinsley attempts to revise the scenarios to resolve them. This may repeat up to several times before the workflow finalizes.

Status Badges

While generation is running, a status badge appears next to the scenario set title on the Scenarios page. It updates in real time to show the current stage:

  • Generating - The initial scenarios are being created.
  • Validating - The relationship review is running.
  • Healing (attempt N of M) - Issues were found and Hinsley is revising the scenarios.
  • Validated (green) - The scenarios passed validation. If healing was needed, the badge notes how many refinement rounds it took.
  • Issues remain (yellow) - Some relationship issues could not be fully resolved after the maximum number of healing attempts. The scenarios are still usable but may benefit from manual review.
  • Failed (red) - The workflow encountered an error. A Retry button appears alongside the badge to restart it.

Viewing the Validation Report

Once generation completes, clicking the green or yellow badge opens the full validation report for that scenario set. The report shows:

  • Pairwise overlaps - Pairs of scenarios that share significant conceptual territory, each rated by severity (high, medium, or low).
  • Coverage gaps - Important futures or outcomes that no scenario in the set addresses.
  • Abstraction & time horizon notes - Whether the scenarios are consistent in scope and time frame.
  • Reasons - A summary explanation of the overall validation outcome.

The report preserves the history of each validation and healing attempt in reverse chronological order, so you can see how the scenarios improved across rounds.