In their 2002 book, Smart Choices: A Practical Guide to Making Better Decisions, Hammond, Keeney and Raifa explain the "PrOACT" model of decision making. As they see it, every good choice has five elements:

Pr = Problem definition

O = Objective

A = Alternatives

C = Consequences

T = Trade-offs

__Problem__

Begin with a written problem definition. What triggered the decision problem? What are the constraints? What related decisions must be made? What is the scope of the decision problem?

__Objective__

List all the concerns to be addressed by this decision. Convert each concern into a specific objective. Separate the means objectives from fundamental objectives. To find fundamental objectives, ask "why is this important" until you have reached the fundamental, underlying objectives. You will use means objectives to find new alternatives, and you will use fundamental objectives to decide between alternatives.

Test your list of objectives; is it mutually exclusive and collectively exhaustive?

__Alternatives__

Generate many alternatives - the more the alternatives, generally, the better the final decision. Ask how each objective might be achieved. Assume each constraint were removed, and then ask what new alternatives that would open up. Think "10X" - assume key variables were ten times their normal range, or 1/10 of their normal value and then ask what new alternatives would open up.

__Consequences__

Build a table of consequences. For each alternative in turn, imagine it is the future and you have chosen that alternative. What were the results? Based on this first cut, eliminate any alternatives that provide inferior futures. Organize the remaining alternatives into a table, showing how each alternative does at meeting each of the fundamental objectives.

__Trade-offs__

Look at what you have left. Eliminate any dominated alternatives; if A is better than B for all fundamental objectives, then eliminate B.

If no clear winner has yet emerged, begin to make what the authors call "even swaps." Let's say you have A and C remaining. A is better than C on some objectives, but C tops A on the others.

We'll use the author's example (pages 87 - 89) to explain even swaps. In this example, there are two fundamental objectives: increase profit and increase market share. Alternative A yields $10 million profits and a 26 percent market share. Alternative C yields $25 million profits, but only a 21 percent market share.

Steps in "even swaps:"

- Determine the change necessary to eliminate an objective from the decision. For example, if you could eliminate the $15 million profit difference then the only objective left would be market share.
- Assess how much change in another objective would compensate for that profit change. Here, the authors assumed a three percent increase in market share would compensate for a profit decrease of $15 million.
- Make the even swap. In this case, the authors deducted $15 million from the projected profits of Alternative C, but added three percent to the market share outcome of Alternative C.
- Eliminate the dominated alternative. In this case, the profit is now the same for each. Market share for Alternative A is 26 percent, while market share for Alternative C is only 24 percent (21 + 3). Eliminate Alternative C.

Your decision will likely require more than one "even swap." If so, just iterate: even swap - eliminate dominated alternative - even swap - eliminate dominated alternative - and so on.