The Decision Making Matrix, Built for Calls That Are Actually Close

Three options have been on the table for a month, each one has a champion in the room, and every meeting ends with someone saying they need more information. A decision making matrix breaks that loop: list the options down one side, the criteria that matter across the top, give each criterion a weight, score every option against each one, multiply and add. The total does not make the call for you. It shows you where the disagreement really is.

That last part is the point most guides skip. The value of the table is rarely the winning number. It is the ten minutes of argument about the weights, which is the argument the team had been avoiding.

This piece covers how to build one in six steps, a worked example, the three ways the matrix misleads, when not to use one at all, and a check to run before you act on the result.

What a Decision Making Matrix Is

The quality body ASQ describes it in one line: “A decision matrix evaluates and prioritizes a list of options and is a decision-making tool.” It goes by a long list of other names, among them Pugh matrix, decision grid, selection matrix, problem matrix and criteria rating form. They are all the same idea.

ASQ recommends it in three situations. When a list of options has to be narrowed to one. When the choice depends on several criteria at once. And after earlier work has cut the options down to a manageable number.

That third condition matters more than it looks. A matrix with nine options and twelve criteria is a spreadsheet nobody trusts. Three to five options and four to six criteria is the useful range, which means the first job is cutting the list.

How to Build a Decision Making Matrix in Six Steps

The procedure below follows ASQ’s, with the order of two steps changed for a reason we explain after.

1. Write the decision as one sentence. “Which city do we open the second office in” is a decision. “Our growth strategy” is not. If the sentence has an “and” in it, you probably have two decisions.

2. List the criteria. What would make an option good? Cost, speed, risk, fit with the people you have, what customers will notice. Ask the people who will live with the outcome, not only the people making it.

3. Cut the criteria to the ones that matter. Merge any two that are really the same thing. “Cost” and “budget impact” are one criterion counted twice, and counting it twice quietly doubles its weight.

4. Set the weights before anyone scores an option. ASQ suggests splitting ten points among the criteria. A criterion worth four points matters twice as much as one worth two. Agree these as a group and write them down.

5. Score each option on each criterion. A one to five scale is enough, with five the best. The guide on untools uses exactly that and notes that any scale works as long as it is consistent.

6. Multiply, add, and discuss. Each score times its weight, summed across the row. ASQ is explicit that the highest score is not automatically the choice. It is the starting point for a better conversation.

The change we make to ASQ’s order is to fix the weights before scoring. If people see the scores first, they adjust the weights, usually without noticing, until their favourite wins. Weights agreed in advance are the main protection against that.

A Worked Example With Three Locations

Here is the shape of a finished matrix. The scenario and the scores are invented to show the method, so replace every number with your own.

A services firm is choosing where to open its second office. The partners agree four criteria and split ten points among them: access to clients gets four, cost of space gets three, hiring pool gets two, and travel time for the founders gets one.

Criterion (weight)	City A	City B	City C
Access to clients (4)	5 (20)	3 (12)	4 (16)
Cost of space (3)	2 (6)	5 (15)	3 (9)
Hiring pool (2)	4 (8)	3 (6)	4 (8)
Founders’ travel (1)	2 (2)	4 (4)	5 (5)
Total	36	37	38

Three options within two points of each other. That is the most common result, and it is useful. It tells you the criteria do not separate the options, so the real decision rests on something the table does not hold yet.

In this example the next question is obvious. City A wins clearly on clients, the criterion the partners said mattered most, and loses on cost. Is the cost difference survivable? That is one specific question with a number attached, and it is a far better meeting than another round of “I just prefer City B”.

A close result is also a reason to check the weights once more. If moving one point from cost to clients flips the winner, the decision is sensitive to that single judgement, and the partners should say out loud whether they really believe clients are worth four.

Three Ways the Matrix Misleads

False precision. A total of 38 against 37 looks like an answer. The scores behind it were guesses on a five-point scale. Any gap smaller than the effect of moving one score by one point is a tie, and should be read as one.

Missing criteria. The table only knows what you told it. The untools guide claims the method “removes the uncertainty and subjectivity” from a decision. It does not. It moves the subjectivity into the choice of criteria and weights, where at least everyone can see it. The criterion that sinks a project is often the one nobody listed: the key person who will leave, the regulator, the founder’s appetite.

Gaming. Once people know a matrix will decide, they score to win. The defence is to score independently, then compare, then discuss the biggest gaps. A partner who gave City B a five on clients while everyone else gave it a two knows something, or wants something, and either way you need to hear it.

When Not to Use a Decision Matrix at All

A matrix suits choices with several options, several criteria and room to be roughly right. It suits some decisions badly.

Decisions you cannot undo deserve more than arithmetic. Our piece on Jeff Bezos’s regret minimization framework covers his split between one-way doors and two-way doors, and a matrix is a two-way-door tool. For selling the company or moving the family, use it to organise the facts and then make the call with more than a total.

Small decisions do not need one. If being wrong costs a week, decide in a minute. The time spent building a table is time not spent on the calls that matter.

And personal decisions sit awkwardly in a grid. Whether to take an executive MBA or accept an offer can be scored, and the scoring helps. Our piece on reading a compensation package is essentially a set of criteria. But the weight you give to family time is not something a colleague can help you set.

Run a Premortem Before You Act on It

Once the matrix has a winner, run one more check. The decision researcher Gary Klein described it in a September 2007 article in Harvard Business Review on performing a project premortem. The team assumes the project they are planning has just failed and generates plausible reasons for its demise.

Applied to a matrix, it takes ten minutes. Imagine it is a year from now and the winning option was a disaster. Write down why. If the reasons point at a criterion you left out or weighted too lightly, go back and fix the table. If they point at something you can manage, write down how.

The premortem works on the matrix’s biggest weakness, the missing criterion, because it asks people to describe failure in their own words rather than in the columns you gave them. It also protects the person who has doubts but did not want to argue against the room.

Where the Matrix Fits in a Working Week

Most executives need one a month at most. Keep a template ready, so building it takes fifteen minutes rather than an afternoon, and put the next one in your productivity plan as a fixed block rather than a meeting that drifts.

Do the scoring in the morning if you can. As we noted in our piece on Jeff Bezos’s morning routine, he puts his hard decisions before lunch, and a tired room scores carelessly.

Pressure distorts it too. A decision forced through after a bad week tends to over-weight speed. Our piece on stress tolerance covers how load changes judgement, and the practical lesson for a matrix is to set the weights on a calm day.

Keep the finished table. A year later it is the best record you have of what you knew and what you assumed, which is the same reason we recommend writing things down within the hour in our piece on how to remember things quickly.

Owners will find the matrix especially useful in the run-up to a sale. Choosing between a broker and a direct approach, or between buyers, is a textbook multi-criteria choice, and our piece on how to sell a small business lists most of the criteria you would need.

The same goes for personal finance, where people often decide between two or three plans on a single number. Our guide to net worth by age argues for judging a balance sheet on more than one measure, and a small matrix is a good way to do it.

The habit of scoring options against stated criteria is an old one. Investors will recognise it from the checklists in some of the Warren Buffett books we covered this week, where the discipline lies in deciding in advance what you are looking for.

Frequently Asked Questions

What is a decision making matrix?

A table that scores a set of options against a set of weighted criteria. Options go down one side, criteria across the top, each criterion gets a weight, each option gets a score on each criterion, and the weighted scores are added up. ASQ notes that the highest total guides the discussion rather than making the choice automatically.

How many criteria should a decision matrix have?

Four to six is usually enough. Fewer and the table misses what matters; more and the weights get thin and the criteria start to overlap. If two criteria measure the same thing, such as cost and budget impact, merge them so they are not counted twice.

What is the difference between a decision matrix and a Pugh matrix?

ASQ lists Pugh matrix as one of several names for the same tool. In a Pugh comparison, each option is scored as better, worse or the same as a chosen baseline option, rather than on an absolute scale. It suits engineering and design choices where one option is the current standard.

The total at the bottom of the table matters less than the argument about the weights that comes before it. Teams that agree the weights in advance and score independently usually find they agreed more than they thought.