Pugh matrix for decisions: comparing options against a baseline

What is a Pugh matrix and how does it compare options?

A Pugh matrix lists criteria in rows and options in columns, with one option chosen as the baseline. Every other option is compared with that baseline, one criterion at a time. The comparison asks a simple question: is this option better than the baseline, the same, or worse? The answer is written as a plus, a zero, or a minus, never as a detailed rating.

The method trades precision for speed. A team that cannot yet agree on a 1 to 10 scale can still agree that option B responds faster than today’s setup. The matrix therefore works well in a first workshop, where it removes weak options in under an hour and leaves a short list worth rating in depth.

How do you choose the baseline option for a Pugh matrix?

The baseline is the reference every other option is judged against. The strongest choice is the current solution, because everyone knows how it performs and the team compares each new idea with something real. When a decision has no current solution, pick the option the team knows best or the one most often proposed. Choosing it deliberately avoids a quiet bias toward the idea that someone already likes.

Write the baseline at the top of the sheet and keep it fixed during the first round. Its own column shows zeros everywhere. A baseline that nobody understands produces guesses in every other column, so describe it in two sentences before scoring. A clear reference makes the plus and minus marks mean the same thing to every reader.

How do you score each option against the baseline in a Pugh matrix?

Score one criterion at a time across all options. Write a plus when the option is clearly better than the baseline on that criterion, a zero when the difference is too small to matter, and a minus when it is clearly worse. Agree in advance what clearly means, for example a difference that a customer or a manager would notice.

Add one short reason beside every plus and minus. Then count the pluses, the minuses, and the zeros for each option, and subtract minuses from pluses to get a net score. The net score is a signal for discussion. A reader who sees a net of plus 2 asks which criteria produced it, and that question drives the next step. Zeros deserve the same attention, since a long run of zeros shows that two options barely differ.

What does a Pugh matrix show for three customer support models?

Take a team that compares three support models with its current in-house team as the baseline, all marks illustrative. The six criteria are cost, response time, quality, scalability, control, and set-up effort. An outsourced partner scores plus, plus, minus, plus, minus, minus, which gives 3 pluses, 3 minuses, and a net of 0.

A hybrid model scores zero, plus, plus, plus, zero, minus, which gives 3 pluses, 1 minus, and a net of plus 2. A chatbot-first model scores plus, plus, minus, plus, zero, minus, which gives 3 pluses, 2 minuses, and a net of plus 1. The hybrid model leads, and the outsourced partner offers no net gain over today’s team.

Should a Pugh matrix for decisions use weights on the criteria?

Weights are an option once the first count is done. Give each criterion a weight from 1 to 3 and multiply it by its mark. In the example, quality carries 3, control carries 2, and the other four criteria carry 1 each. The weighted total then rewards the options that win on the criteria that matter most.

The outsourced partner falls to minus 3, since its minus on quality costs 3 and its minus on control costs 2. The chatbot-first model reaches minus 1. The hybrid model rises to plus 4, because its plus on quality adds 3. The weighted view widens the gap, and a reader sees that the quality criterion drives the whole ranking.

How do you iterate a Pugh matrix to narrow the options?

Run a second round with the best option of round one as the new baseline. In the example, the hybrid model becomes the reference, and the other two options are compared with it. Options that still score minus on most criteria leave the list, and the team keeps the ones that beat the leader on at least one criterion.

Iteration also invites combination. If the chatbot-first model wins on cost and the hybrid model wins on quality, a third design can take the cost advantage of the first into the second. The team then writes the new option into the matrix and scores it. Two or three rounds usually leave one or two options ready for a full weighted rating.

When does a Pugh matrix need a decision engine behind it?

A Pugh matrix hides the size of each difference. A minus for a 2 percent cost gap and a minus for a 40 percent gap look identical, and the baseline itself shapes every result. Use the matrix to screen options, then move the short list to a weighted matrix or to expected value, where magnitudes and probabilities enter the calculation.

The EQOS Asia decision engine goes further: its Audit pillar measures a situation on 9 structural dimensions and returns a φ(S) score between 0 and 1 (EQOS Asia method page, as of October 2026). The documented method fixes the dimensions in advance, and the simulator lets a reader try a first score on the shortlisted option.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile