Organizations inform people, solve problems, and make decisions. The activities overlap, but they produce different things. Analysis can explain a condition. Problem solving can apply a known method. A decision commits people and resources to one course of action while material facts remain uncertain.
Confusing those activities leads to a common failure: a team solves the problem it was given without establishing that it is the problem the organization needs to decide.
From an Issue to a Decision
Consider a large customer whose satisfaction and margin are both declining.
- Issue: The customer relationship needs attention.
- Question: Should the company renegotiate, restructure the service, or exit?
- Hypothesis: A specified set of new terms will retain the customer and restore margin above a stated threshold.
- Assumptions: The customer will accept those terms, service costs will behave as estimated, and the contract permits the change.
- Scenarios: The customer accepts, rejects, delays, or accepts while demand changes.
- Analysis: Estimate the outcome distribution for each option, identify the assumptions that drive it, and examine the losses in adverse cases.
- Decision: Choose an option, assign an owner, and record the evidence that would trigger review.
The first hypothesis, “renegotiation might save the relationship,” is too vague to test. The useful work begins when the team names the terms, threshold, customer alternatives, contractual limits, and timing.
Uncertainty should remain visible in the analysis. If demand, cost, or timing is uncertain, a single forecast hides information the decision-maker needs. A range, scenario set, or probability distribution preserves it. Sensitivity analysis identifies which uncertain inputs can change the choice. Stress tests ask which combinations produce an unacceptable result.
Monte Carlo simulation is useful when several quantified uncertainties interact and repeated sampling represents the mechanism credibly. It is not mandatory. A decision with sparse evidence may be better served by a few explicit scenarios, bounds, and a staged commitment. Use the simplest method that represents the uncertainties capable of changing the decision.
Nine Dimensions
The following dimensions determine the analysis and controls a decision needs.
| Dimension | Question to ask | Common error |
|---|---|---|
| Frequency and consequence | How often will we make this choice, and what does one error cost? | Treating a recurring budget process as routine even though it allocates strategy |
| Reversibility | What can be undone, until when, at what cost, and for whom? | Calling a reorganization reversible after people and knowledge have left |
| Decomposability | Can sub-decisions be evaluated independently? | Scoring projects separately when they compete for the same team or capital |
| Ambiguity | Do the outcomes share a measure, or must we judge among different values? | Pretending team health and individual output reduce to one objective function |
| Time horizon | When do costs, benefits, and evidence arrive? | Evaluating an acquisition only on its first-year integration |
| Stakeholders | Who benefits, pays, supplies information, or holds a veto? | Inviting only the function that will execute the choice |
| Information | What is known, missing, costly to reveal, or likely to arrive too late? | Waiting for another quarter without pricing the lost option |
| Interdependence | Which other decisions share assumptions, resources, or sequence? | Changing growth targets without revisiting capital and hiring commitments |
| Precedent | What will this choice make easier, expected, or politically protected next time? | Treating the first discount or policy exception as isolated |
The dimensions interact. An exit may be legally reversible but reputationally costly. A technology migration may be sound in isolation but conflict with customer commitments. A small commercial exception may become the reference for every later negotiation.
The spectacular decisions, such as an acquisition, IPO, or market exit, are easy to recognize as consequential. More damage often accumulates in ordinary budget reviews, renewals, staffing choices, and roadmap meetings whose process understates their precedent or interdependence.
Six Recurring Failure Modes
Default: A product line keeps receiving last year’s budget because stopping requires a visible decision. Nobody compares continuation with exit until a crisis makes exit more expensive.
Precedent: An introductory price wins one account and becomes the reference for the next ten. A tactical exception then fixes the economics of a strategic customer segment.
Wrong model: The calculations are correct, but the causal model is not. A company models competition through market share when switching costs or network effects determine the outcome. Internal consistency cannot rescue a false premise.
Wrong level: Each business unit rationally protects its P&L, so a cross-unit investment that creates portfolio value receives no sponsor. The incentive system rejects a choice that nobody evaluates for the company as a whole.
Delay: Market entry has positive expected value at the time of analysis, but four months of refinement allow a competitor, hiring market, or rule change to remove the opportunity. The delay was part of the decision and should have been priced.
Narrow participation: A technical team chooses a migration without the commercial owner who knows the revenue calendar, the lawyer who knows the service obligations, or the operators who know where the undocumented behavior sits. The answer fits the frame; the frame excludes a material constraint.
The Questions Recur
Strategy consists partly of decisions that must be reopened as conditions change.
| Area | Recurring questions |
|---|---|
| Market and product | Which customers and markets do we serve? What do we build, buy, or obtain through a partner? How do we price it? |
| Capital | Where does the next dollar go? What mixture of cash flow, debt, and equity supports the plan? What expectations do the terms create? |
| People and organization | Which structure carries the required information and authority? What must remain an internal capability? |
| Technology | Which platforms constrain future choices? When does migration cost less than continued maintenance? |
| Risk and governance | Which risks do we accept, reduce, transfer, or avoid? Who can authorize each exposure? |
| Growth | Do we grow internally, acquire, partner, divest, enter public markets, or leave them? |
| Stopping | Which commitments have lost their rationale? What evidence distinguishes patience from delay? |
An executive team can review the list and ask four questions of each active decision: What did we decide? On what evidence and assumptions? Who owns the commitment? What new evidence requires reconsideration? A missing answer identifies a decision being made through habit or local incentives.
What Decision Support Should Provide
A recommendation asks a decision-maker to approve a conclusion. Decision support exposes enough of the analysis for that person to exercise judgment. It should provide:
- the issue, question, alternatives, and scope;
- material evidence for and against each alternative;
- assumptions with owners, sources, ranges, and review dates;
- scenarios, outcome distributions, sensitivities, and stress losses where appropriate;
- dependencies on other active decisions;
- implementation, reversal, and stopping conditions; and
- a record of the choice and the information available when it was made.
Different tools have different evidentiary roles. Code should perform financial algebra, simulation, optimization, and consistency checks when their inputs can be specified. A model can search sources, compare documents, identify missing assumptions, draft an explanation, or challenge a causal claim. Another model can propose a different critique. Agreement among models does not validate the result; primary evidence, executable checks, and accountable domain review do.
Better tools reduce the cost of preparing this material. They do not choose the objective, decide which tradeoffs are acceptable, or accept responsibility for the commitment. The person with authority still decides.
The standard is simple: preserve the uncertainty, expose the assumptions, connect the decision to its dependencies, and record what would make the organization change course.