Scenario Analysis Framework

Definition

Scenario analysis is a structured method for evaluating how different sets of assumptions — representing distinct future states of the world — affect a financial model's outputs. Unlike sensitivity analysis (which varies one input at a time), scenario analysis changes multiple correlated assumptions simultaneously to model coherent alternative futures, typically organized as base case (most likely), bull case (optimistic), and bear case (pessimistic). [src1]

Key Properties

• Standard structure: Three scenarios — base (50-60% probability), bull (20-30%), bear (15-25%) [src1]
• Implementation: Scenario toggle cell using INDEX/MATCH, OFFSET, or CHOOSE functions [src2]
• Scope of variation: Revenue, margins, customer metrics, pricing, macro assumptions all change together [src3]
• Output: Scenario-specific financials plus probability-weighted expected values [src1]
• Internal consistency: Each scenario must represent a plausible, self-consistent world state [src2]

Constraints

• Internal consistency is mandatory: A bear case with pessimistic revenue but optimistic costs is a modeling error. [src2]
• Three scenarios may be insufficient: Complex businesses miss compound risks with only three cases. [src3]
• Probability weights are subjective: Assigning precise probabilities creates false precision. [src1]
• Discrete, not continuous: Tests specific states — use sensitivity analysis or Monte Carlo for continuous ranges. [src4]
• Requires existing base model: Scenario analysis is a layer, not a standalone framework. [src2]

Framework Selection Decision Tree

START — User needs to test model under different conditions
├── What kind of uncertainty?
│   ├── Multiple correlated assumptions → Scenario Analysis (this unit)
│   ├── Individual variables one at a time → Sensitivity Analysis
│   ├── Probability distributions, many trials → Monte Carlo Simulation
│   └── Binary risk events → Stress Testing
├── How many scenarios needed?
│   ├── 2-5 discrete states → Standard scenario analysis
│   ├── 10+ with probability weighting → Scenario tree
│   └── Continuous distribution → Monte Carlo
└── Does a base model exist?
    ├── YES → Layer scenarios on top
    └── NO → Build base model first

Application Checklist

Step 1: Define scenario narratives

• Inputs needed: Key uncertainties (market, competitive, regulatory, macro)
• Output: 3-5 named scenarios with 1-2 sentence narratives
• Constraint: Each narrative must be internally consistent [src2]

Step 2: Map assumptions to each scenario

• Inputs needed: Base assumptions, narrative-driven adjustments
• Output: Assumption matrix (variables x scenarios)
• Constraint: Do not change only revenue — costs, churn, and capital must shift consistently [src1]

Step 3: Build the scenario toggle mechanism

• Inputs needed: Assumption matrix, base model
• Output: Single toggle cell switching all assumptions via INDEX/CHOOSE
• Constraint: Toggle must change every linked assumption — partial switches break consistency [src2]

Step 4: Generate outputs and probability-weighted values

• Inputs needed: Scenario-specific statements, probability weights
• Output: Summary table per scenario plus weighted expected value
• Constraint: If weighted output differs from base by 20%+, address the implied skew [src1]

Anti-Patterns

Wrong: Bear case is "base case minus 10%"

Mechanically reducing base revenue by 10% without adjusting costs or churn produces an unrealistically mild downside. [src1]

Correct: Bear case tells a coherent downside story

Start with a narrative ("recession causes 2x churn, 50% slower sales") and derive all assumptions from that story. [src2]

Wrong: Only varying revenue across scenarios

Changing only the top line while holding all else constant is sensitivity analysis on one variable, not scenario analysis. [src3]

Correct: Varying all correlated assumptions together

In a recession scenario, adjust revenue AND churn AND sales cycle AND payment terms AND hiring simultaneously. [src2]

Wrong: Presenting probability-weighted values as predictions

Reporting a weighted average valuation as "the expected value" obscures the wide range and subjective weights. [src4]

Correct: Presenting the range alongside the weighted average

Show the full range (bear to bull) with the weighted average as one reference point. Emphasize assumptions, not the single number. [src1]

Common Misconceptions

Misconception: Scenario analysis and sensitivity analysis are the same thing.
Reality: Sensitivity analysis varies one input at a time. Scenario analysis changes multiple correlated inputs simultaneously. They are complementary, not interchangeable. [src1]

Misconception: Three scenarios are always sufficient.
Reality: Complex businesses with multiple independent uncertainties may need scenario trees or Monte Carlo. A startup facing market AND regulatory risk may need 9 scenarios. [src3]

Misconception: The base case is the "most likely" outcome.
Reality: The base case is the central expectation, but any single scenario has low probability of occurring exactly as modeled. It is a reference point, not a prediction. [src2]

Comparison with Similar Concepts

When This Matters

Fetch this when a user asks about building financial scenarios, creating base/bull/bear cases, presenting scenario-based projections, or structuring a scenario manager in a financial model.