Learn
How decision analysis structures a hard choice
Decision analysis is a formal way to choose under uncertainty. You lay out the options, put probabilities on the events that matter, value each outcome, and choose the path with the best balance of reward and risk. Ronald Howard of Stanford gave the field its name in 1966, and its tools remain the quiet backbone of serious choices in energy, medicine, and finance.
The tools of the craft
Three tools do most of the work. A decision tree draws each choice as a branch and each uncertain event as a fork, with a probability on every fork. Expected value multiplies each outcome by its probability and adds the results, so that one branch can be compared with its rivals in a single figure. Sensitivity analysis then asks how far each input can move before the preferred branch changes.
A fourth tool matters when outcomes are not all money: a weighted set of criteria, which lets a team compare speed, quality, and risk on one scale. Most real decisions use some mixture of the four.
Running an analysis
Write down the decision and the date by which it must be made. List the options, including waiting. For each option, name the two or three uncertain events that drive its result, and give each a range rather than one figure: a low case, a likely case, and a high case. Work out the expected value of each option, then test which input, moved across its range, changes the answer.
That final test is the heart of the method. It tells you where to spend the next week of research, because only the inputs that can reverse the ranking deserve it.
The mistake that undoes it
Treating the first model as the truth. A tree built in an afternoon inherits every guess its author made, and a precise figure such as $1.84 million carries no more certainty than the rough probability inside it. The second mistake is quieter: analyzing only the options someone happened to bring to the meeting, when the best path was never written down.
How Argumont carries the method
In a Decision Room, criteria carry weights that total 100 before any option is scored, and the engine works the sums from your own figures exactly. Each option has a Champion, and the Red Teamer runs the stress cases a sensitivity test would ask about: the lower price, the later launch, the lost customer. The Strategist looks for a path the tree left out.
The Verdict states which option leads, what it rests on, and the conditions that would reverse it. The Human Decision remains yours and is recorded beside it.
Launch now or wait, by expected value
A software company can launch in March or wait until June for a feature its largest prospect has asked for. Launching now earns about $600,000 in the first year if the prospect signs anyway, which the team puts at 40 percent, and $250,000 if it does not. Waiting shortens the year, but the team believes the prospect then signs with 80 percent likelihood, for $520,000, or $180,000 if it still declines. Launching now is worth 0.4 × $600,000 plus 0.6 × $250,000, or $390,000. Waiting is worth 0.8 × $520,000 plus 0.2 × $180,000, or $452,000. Waiting leads, but only while the 80 percent holds; at about 62 percent the two paths are even, so that single estimate is the one to check before June.
Questions
- Is decision analysis the same as a decision matrix?
- A matrix is one of its tools. Decision analysis also handles probabilities and outcomes that unfold over time, which a matrix on its own cannot show.
- Do I need exact probabilities?
- Rarely. Ranges and rough estimates are enough to run the method, and the sensitivity test tells you which estimates are worth refining before you commit.
Bring Argumont a Decision worth challenging.
Start with a paid 60 day pilot on your own real Decisions.