Decision Matrix: How to Use a Weighted Pugh Matrix With Template

By Brian Tr · Updated Oct 10, 2026

A decision matrix is a grid that scores several options against the same set of criteria so you can compare them side by side. Criteria sit in rows, options sit in columns. In a Pugh matrix you rate each option against a baseline option. In a weighted scoring matrix you score each option directly and multiply by how much each criterion matters.

This guide covers both methods, how to pick criteria and weights, which scoring scale to use, how to stress-test the result, and what to do when two options tie. It also includes a worksheet you can paste into a spreadsheet.

Key Facts

• What it does: turns a vague "which one?" debate into explicit criteria, scores, and a ranked result the team can inspect and challenge.
• Two main variants: the Pugh (datum) matrix compares every option to one reference option using +, S, and -. The weighted scoring matrix gives each option a number per criterion and multiplies by a weight.
• Origin of the Pugh method: Stuart Pugh laid it out in his 1991 book Total Design as a way to move from many concepts to a few strong ones. The ESD.33 course notes from MIT OpenCourseWare summarize the procedure.
• Best used for: vendor selection, location choices, product concepts, hiring finalists, and investment shortlists where several factors pull in different directions.
• Biggest risk: false precision. A tidy total of 3.8 versus 3.6 can hide a guess about weights.
• Always do: a sensitivity check. Change the weights and see whether the winner changes.

What is a decision matrix?

A decision matrix (also called a decision grid, selection matrix, or criteria-based scoring matrix) is a structured way to compare options when no single factor decides the matter. Price might favor option B, risk might favor option C, and speed might favor option A. The matrix forces you to say how much each factor matters and how each option performs on it.

Government decision-support guidance describes the same idea under the name multi-criteria analysis. The UK's Multi-criteria analysis manual lays out the sequence: set the decision context, list the options, identify criteria, describe or score how each option performs, assign weights, combine weights and scores, examine the results, and run a sensitivity analysis. Every practical decision matrix is some version of that sequence.

The matrix doesn't make the decision. It makes your reasoning visible. When someone disagrees with the outcome, you can point to a specific weight or score and argue about that, instead of arguing about gut feeling.

Pugh matrix vs weighted scoring matrix

People often use "decision matrix" and "Pugh matrix" as if they were the same thing. They aren't. They answer slightly different questions.

The Pugh (datum) matrix

In a Pugh matrix you pick one option as the datum, a reference everything else is compared against. It's usually the current solution or the concept the team understands best. For each criterion, you mark every other option as better (+), about the same (S), or worse (-) than the datum. Then you count the pluses and minuses.

The MIT course notes describe the first run like this: generate multiple solutions, choose and clarify criteria, choose a datum, insert the comparisons, then look at the strongest concepts and try to fix their negatives by borrowing from complementary concepts. Weak concepts get improved or eliminated. A second run follows with the strongest survivors, expanded criteria, and often a new datum.

The point of the Pugh method isn't to crown a winner in one pass. It's controlled convergence: eliminate the weakest ideas while looking for hybrids that combine strengths. Pugh was also wary of treating the counts as math. The same course notes quote his view that the method avoids "the rigidity and false confidence of rating/weighting matrices" and that its numbers are guidance only, not something to sum algebraically.

The weighted scoring matrix

A weighted scoring matrix goes the other way. Each criterion gets a weight, each option gets a score on each criterion, and you multiply and add. The result is a single number per option, which is easy to rank and easy to present.

That's also why it's easy to misuse. The UK manual notes that simple weighted averages let a low score on one criterion be offset by a high score on another, and that the method depends on the criteria being reasonably independent of one another. If a hard minimum matters (a safety standard, a legal requirement), don't let it be averaged away. Treat it as a pass/fail gate before the scoring starts.

The hybrid most teams actually use

Many teams combine the two: a Pugh-style first pass to cut a long list down to three or four, then a weighted score on the survivors. That's the "weighted Pugh matrix" people search for. It keeps the discussion honest early, when you have many vague options, and adds numeric ranking later, when the differences are smaller.

Decision matrix vs other decision tools

A decision matrix isn't always the right tool. Here's how it sits next to the usual alternatives.

A useful sequence: use SWOT or gap analysis to understand the situation, generate options, screen them with a Pugh pass, then run a weighted matrix on the finalists. If the money is the main story, put a cost-benefit analysis alongside it, or make "net cost" one of your criteria. The Eisenhower matrix belongs somewhere else entirely. It prioritizes tasks you've already committed to, so it can't tell you which option to choose.

How to build a weighted decision matrix

Step 1: Define the decision and list the options

Write the decision as a question with a deadline: "Which vendor do we sign for the 2027 warehouse contract by the end of Q1?" Then list the real, feasible options. Include "do nothing" or "keep the current approach" when that's a genuine possibility. Three to six options is the sweet spot. More than that, and you should run a Pugh-style screening pass first.

Step 2: Set pass/fail gates

Some requirements aren't tradeable. An option that fails a legal, security, or budget ceiling requirement shouldn't get a score at all. Remove those options before you build the grid, so the weighted math doesn't quietly let a strong score elsewhere cover for a disqualifying flaw.

Step 3: Choose the criteria

Pick the factors that actually separate the options. Aim for five to eight. Fewer than that and you're probably ignoring something. Many more and the weights get diluted, which makes the total look stable when it isn't.

Three rules keep criteria clean:

• Make each one measurable or at least describable. "Quality" is vague. "Defect rate on the last two pilot shipments" is something two people can score the same way.
• Avoid overlap. The UK manual warns about double counting: if the same underlying effect appears under two criteria, it quietly gets more weight than you meant to give it. "Cost" and "total budget impact" usually describe the same thing.
• Use the same direction for all. Decide that a higher score is always better. For things like cost or risk, define the scale so that 5 means lowest cost or lowest risk.

Step 4: Assign weights

Weights express how much each criterion matters to this decision. A common approach is to give the criteria 100 points to split. If cost matters twice as much as onboarding effort, it gets roughly twice the points.

Here's the subtlety the UK manual stresses, using a method called swing weighting. A weight should reflect two things: how much you care about a criterion, and how far apart the options actually are on it. If every option is equally safe, safety shouldn't carry a large weight in this particular comparison, even if safety matters enormously in general. If the options differ by a huge margin on maintenance cost, that criterion deserves real weight even if it feels like a secondary concern.

Practical tips:

• Have each decision-maker assign weights separately, then compare. Differences in weights show where the real disagreement is.
• Fix the weights before you see the option scores. Otherwise people tune the weights to protect a favorite.
• Write one sentence explaining the top two weights.

Step 5: Pick a scoring scale

Common choices:

Whatever you choose, write an anchor for the top and bottom of each criterion. For example, for delivery speed: 1 = more than 10 days, 5 = under 3 days. Anchors keep two scorers from meaning different things by "4".

Step 6: Score each option

Score each option on each criterion, ideally with evidence: quotes, pilot results, reference calls, test data. Have more than one person score independently where you can, and discuss the biggest differences. When scorers disagree, that's useful information. Don't average it away. Find out what one of them knows that the other doesn't.

Step 7: Calculate and rank

Multiply each score by its weight, then add the results for each option. If your weights sum to 100 and your scale runs 1 to 5, divide the total by 100 to get a weighted average on the same 1 to 5 scale.

Step 8: Run a sensitivity check

This is the step most teams skip, and the one that matters most. The UK manual lists it as the final step in the process: change the weights and scores and see whether the order changes.

Try at least three tests:

• Shift the top weight up and down by 10 points.
• Swap in a different stakeholder's weights.
• Change one uncertain score by one point and see if the winner flips.

If the winner stays the winner across all of them, you can be confident. If it flips easily, the decision rests on a judgment call, and you should talk about that judgment directly rather than hiding it inside the arithmetic.

Step 9: Decide, and record why

Treat the result as input, not as a verdict. Check it against your gut. If the matrix says B but everyone in the room feels uneasy, something is missing from the criteria. Add it and rerun. Then write down the options, weights, scores, and the reason for the final choice so the decision can be revisited later.

Worked example: choosing a logistics supplier (hypothetical)

Suppose a company must choose among three logistics suppliers: A (the current one), B, and C. The team agrees on five criteria and splits 100 points of weight. Scores run 1 to 5, where 5 is best. All numbers here are invented for illustration.

The arithmetic for Supplier C: (3 x 30) + (5 x 25) + (4 x 20) + (5 x 15) + (3 x 10) = 90 + 125 + 80 + 75 + 30 = 400, which is 4.00 on the 5-point scale. Supplier C leads.

What a Pugh pass says about the same options

Now treat Supplier A as the datum and mark B and C against it.

On the Pugh count, B and C tie. The weighted matrix breaks that tie because it knows that speed and risk carry more weight than onboarding. That's a fair illustration of why the two methods complement each other: Pugh shows where each option is stronger or weaker, and weighting says how much those differences should count.

The sensitivity check

Now stress-test it. Suppose the finance lead argues that cost should dominate, with weights of cost 45, speed 15, quality 20, risk 10, and onboarding 10. Rerunning the totals gives Supplier A 3.75, Supplier B 4.05, and Supplier C 3.70. Supplier B now wins.

That flip is the real finding. C wins if the team values speed and low risk; B wins if cost is the overriding concern. The useful conversation isn't "who has the highest score?" but "do we value cost that heavily?" The matrix has turned an argument about suppliers into an argument about priorities, which is easier to settle.

What to do when options tie

Ties happen, especially on a 1 to 3 or 1 to 5 scale. Before you flip a coin:

1. Check the margin. Differences of a few percent are within the noise of your own scoring. Treat close scores as a tie.
2. Run the sensitivity tests. If one option wins in most scenarios, prefer it.
3. Look at the biggest single gap. An option that is excellent on the most heavily weighted criterion and mediocre elsewhere is a different bet from one that is solid everywhere.
4. Add a tie-breaking criterion you hadn't considered, such as reversibility, learning value, or strategic fit, then say so openly.
5. Look for a hybrid. This is the Pugh instinct: combine the strongest elements of the tied options.
6. Pilot both. If the decision is reversible, a small test beats more debate.

Free decision matrix template

Copy this into a spreadsheet. Replace the example criteria with your own, fill in the weights (they should sum to 100), score each option, and let the formulas do the rest.

Add three extra rows below the table:

• Pass/fail gates: the requirements an option must meet to be scored at all.
• Sensitivity test: the alternate weights you tried and which option won each time.
• Decision and reason: the choice, the date, and one sentence on why.

For a Pugh version, replace the score columns with +, S, or - against a named datum, and count the totals at the bottom without weighting them.

Decision matrix examples by business function

• Procurement: compare suppliers on unit cost, delivery reliability, quality, financial stability, and contract flexibility. For a more specialized view of supplier risk and leverage, see the Kraljic matrix.
• Product: rate feature ideas on customer impact, effort, strategic fit, and risk. Use a Pugh pass first to cut a long backlog down.
• Operations: choose between sites, tools, or process designs on cost, capacity, lead time, and ease of change.
• Finance: weigh investment options on return, payback period, risk, and capital required, alongside a cost-benefit analysis.
• HR: rank finalist candidates on skills, experience, culture contribution, and growth potential, with the criteria fixed before interviews begin. Related reading: decision-making as a competency.
• Strategy: compare market entry options, partnerships, or acquisition targets on market size, fit, risk, and speed to revenue.

Decision matrix dos and don'ts

Do

• Agree on the criteria and weights before anyone scores an option.
• Write anchors for the top and bottom of each scale.
• Keep pass/fail requirements outside the weighted math.
• Run a sensitivity check and report what flipped.
• Keep a record of the weights and the reasoning.
• Use the result to start a conversation, not end one.

Don't

• Don't add criteria after you've seen the totals, just to rescue a favorite.
• Don't count the same factor twice under different names.
• Don't use ten or more criteria unless you have a real reason. Weights get thin and the result gets mushy.
• Don't treat 3.8 versus 3.7 as a difference. Treat it as a tie.
• Don't ignore a strong gut reaction. It often points to a missing criterion.
• Don't rely on a numeric score for decisions where criteria interact strongly. The MIT course notes show a case where a simple count can rank combinations of attributes badly, because the best overall design isn't always the sum of the best individual attributes.

Limitations of the decision matrix

A decision matrix is a thinking aid, not an oracle. Weights are subjective. Scores depend on who is scoring. Groups don't have a single "correct" preference to uncover, and the MIT notes make the point that there is no perfect way to combine individual preferences into an optimal group choice. The matrix gets you transparency and structure, and that's valuable. It doesn't get you objectivity for free.

Used well, it surfaces assumptions, shows which disagreements matter, and leaves a record. Used badly, it dresses a preference up as a calculation.

Keep your decisions defensible

Build the matrix, stress-test it, and write down why you chose what you chose. If you're deciding what to analyze before you score anything, start with a SWOT analysis or a gap analysis, then bring the finalists here.