Decision-making frameworks
Most decision frameworks are sorting tools. They take a choice you're already carrying around and give it a shape — columns, weights, a horizon, a devil's advocate — so that what you're weighing sits outside your head where you can look at it. That's worth doing, and it's narrower than the way these things are usually sold, because a framework can only work on the information you walk in with.
Here are six that earn the time, what each one is good at, and where each one gives out.
The pros and cons list
Benjamin Franklin described the version worth using in a 1772 letter to Joseph Priestley, the English chemist, and called it moral algebra. Write the two columns over several days rather than one sitting, so that considerations arrive as they occur to you. Then cross out items of roughly equal weight from opposite columns, a pro against a con, until one side is empty. Whatever survives is the answer.
The crossing-out is the part that usually gets dropped, and it is the part doing the work. A plain two-column list has no weights, so one decisive objection sits at the same size as five conveniences, and you can lengthen either column at will. If you already know which way you're leaning, an unweighted list will ratify it.
Good for: getting the considerations out of your head, early, when you're still discovering what you're weighing.
The weighted decision matrix
Options as rows, criteria as columns, a weight on each criterion, a score in each cell, multiply and sum. It's the pros and cons list with the weights made explicit.
The weights are the whole answer, and you set them, which means the method can turn a preference you already had into arithmetic that looks impartial. There's one habit that mostly fixes this: set the weights before you look at the options, and write down why each one got the number it did. If the winner then surprises you and you find yourself reaching to adjust a weight, the reach is the useful information, not the score.
Good for: choices with several real criteria and no dominant one — two job offers where money, commute and the actual work all matter.
Expected value
Multiply each outcome by its probability, add them up, pick the highest number. It is the correct calculation for anything you'll face repeatedly, rather than an aid to judgement.
It breaks on one-shot decisions, where the average is a number you will never experience, because you get a single draw. And it says nothing about ruin: an option with the best expected value and a 5% chance of ending the whole enterprise is not the option, and the arithmetic won't tell you that.
Good for: repeated decisions with real numbers attached, and as a sanity check on any decision where you've been reasoning in adjectives.
The outside view
Rather than reasoning forward from the specifics of your case, ask what happened to the last hundred people in roughly your position, and start there.
The psychologists Daniel Kahneman and Dan Lovallo, writing in Management Science in 1993, separated the two modes: the inside view builds a forecast from the details of this case and reliably produces an optimistic one, while the outside view starts from the record of the class the case belongs to. Bent Flyvbjerg, a Danish economic geographer who has spent decades collecting cost and schedule data on large infrastructure projects, found overruns to be the normal outcome rather than the exception, and the honest forecasts came from teams who started with comparable projects instead of their own plan.
It depends entirely on the reference class you choose. If comparable cases are genuinely thin, there's nothing to stand on, and where several classes fit, everyone picks the flattering one.
Good for: any estimate of cost, duration, or success rate — the three things the inside view gets wrong in the same direction every time.
The premortem
Before committing, assume it is 12 months from now and the decision failed. Everyone writes down why, independently, before anyone speaks. Gary Klein, a research psychologist who studies how experts decide under pressure, described the exercise in Harvard Business Review in 2007; his argument for it rests on a 1989 experiment by Mitchell, Russo and Pennington, in which treating a future event as though it had already happened raised people's ability to generate concrete reasons for it by about 30%.
It surfaces the objections that momentum and politeness suppress, which is most of them. What it can't do is choose: a premortem criticizes one option and doesn't rank two, so it belongs after you've provisionally decided, not instead of deciding.
Good for: commitments that are expensive to unwind, especially ones a group has already grown attached to.
Regret minimization
Project yourself to 80, look back at the choice, and pick the option you'd regret less. Jeff Bezos has described using this to decide whether to leave a finance job and start Amazon. Suzy Welch's 10/10/10 is the same move at shorter range: how will I feel about this in 10 minutes, in 10 months, in 10 years.
Both work by moving you to a vantage point where whatever feeling is currently distorting the decision has expired — an insult, a bad quarter, a flattering offer that has to be answered by Friday. That's what both are for, and they're weak on anything outside it.
The bias baked in is worth knowing. The regret we picture at 80 is mostly the regret of not having tried, so regret minimization leans toward the bold option almost regardless of the specifics. For a reversible decision that's a reasonable default. For one with a downside you can't walk back, it's a thumb on the scale.
Good for: decisions being made inside a strong feeling, where the feeling will not survive the month.
The question none of them answer
Every framework here ends at the moment you choose. None of them produce feedback, which means you can run a careful matrix, weight it honestly, pick well, and still never find out whether your weights or your probabilities were any good.
The reason isn't laziness about following up. It's that by the time the outcome arrives you no longer have the original numbers to compare it against; you have a recollection of them, and a recollection formed after the fact reads the outcome back into the estimate. Hindsight bias has been reproduced on students, physicians and judges since the 1970s, and it lands hardest on exactly the people trying to learn from their own track record.
The missing step is cheap and it happens before the outcome exists: alongside whatever framework you used, write down what you expect to happen and how sure you are, as a number. Then the framework has something it can be scored against later, and after twenty or so scored decisions you can see which of these tools was actually helping you and which one you were using to feel decisive. That's a decision journal, and there's a template with the fields.
Full disclosure: I make Reckon, an iOS decision journal that keeps the prediction, the confidence number and the scoring, and takes no view at all on how you arrived at the decision — matrix, premortem, coin flip, gut. One-time, no subscription. The frameworks above are all free and most of them work on paper; the record of how they went is the part that needs somewhere to live.