Decision analysis is a structured, often quantitative, method for comparing choices when outcomes are uncertain and multiple objectives may conflict. It brings together data, probability estimates, and a clear statement of goals so decision‑makers can compare alternatives and see the trade‑offs. The field was named and developed in the 1960s (Ronald A. Howard is commonly credited with originating the term) and is applied in areas such as capital budgeting, operations, marketing, strategy, and risk management.
Core concepts (short definitions)
– Decision tree: a graphical map of choices (decision nodes), uncertain events (chance nodes), and outcomes (terminal nodes) that lays out possible paths and payoffs.
– Influence diagram: a compact graphical representation showing decisions, uncertainties, and objectives and how they relate.
– Probability (of an uncertainty): a numeric estimate of how likely a particular event or state will occur.
– Expected value (EV): the probability‑weighted average of possible monetary outcomes. EV = sum of (probability × payoff).
– Utility function: a mapping from outcomes (often money) to a measure of value that reflects the decision‑maker’s risk preferences (risk‑neutral, risk‑averse, or risk‑seeking).
– Analysis paralysis: the risk of over‑analyzing to the point that an actionable decision is delayed or never taken.
Why use decision analysis?
– Forces clarity about objectives and what “success” means.
– Structures complex problems involving multiple uncertain inputs.
– Makes assumptions explicit and facilitates sensitivity testing.
– Produces a documented rationale for a choice, useful for review and learning.
Typical tools and inputs
– Decision trees or influence diagrams to visualize choices.
– Probability estimates for uncertain events (from data, experts, or scenarios).
– Cost and revenue estimates or other outcome metrics.
– Utility functions when value is not linear in money (e.g., when risk aversion matters).
– Simulation models and software to explore many scenarios and parameter combinations.
Practical checklist for performing decision analysis
1. Define the decision and the primary objective(s).
2. Enumerate the feasible alternatives.
3. Identify key uncertainties that affect outcomes.
4. Assign probabilities or scenarios to each uncertainty (document source and confidence).
5. Quantify payoffs for each outcome (costs, revenues, other metrics).
6. Build a decision tree or influence diagram that links decisions, uncertainties, and outcomes.
7. Calculate expected values or expected utilities for each decision path.
8. Run sensitivity analysis (change key probabilities or payoffs to see effect on the recommendation).
9. Consider non‑quantifiable factors (reputation, regulatory risk, strategic options).
10. Choose a decision, record assumptions, and plan monitoring/updates.
Worked numeric example (small, step‑by‑step)
Scenario (adapted from the patent example): A firm owns a patent and has two options:
A) Sell the patent now for $2,000,000 (certain).
B) Produce the product in‑house. Producing requires $1,000,000 in upfront costs. Two years of high sales are possible; management estimates a 60% chance of strong sales that yield $6,000,000 in revenue (total) and a 40% chance of poor sales that produce no revenue. Ignore time value of money for simplicity.
Step 1 — Compute payoffs if the firm builds:
– If success (probability 0.60): profit = revenue − cost = $6,000,000 − $1,000,000 = $5,000,000.
– If failure (probability 0.40): profit = −$1,000,000 (sunk development cost; no offsetting revenue).
Step 2 — Compute expected value of building:
EV_build = 0.60 × $5,000,000 + 0.40 × (−$1,000,000)
EV_build = $3,000,000 + (−$400,000) = $2,600,000
Step 3 — Compare to selling:
– EV_sell = $2,000,000 (certain)
Decision under expected‑value (risk‑neutral) criterion:
– EV_build ($2,600,000) > EV_sell ($2,000,000), so building has the higher expected monetary value.
Important caveats (from this example)
– This conclusion assumes risk neutrality (decision‑maker