Pugh Matrix: Turn a Diagnosed Cause Into a Change Worth Piloting

The Pugh Matrix compares alternative solutions against a baseline using plus, zero and minus scoring across criteria to identify the strongest concept. The baseline is the way the process runs today. Every candidate is judged against it, criterion by criterion.

You reach for it at a specific moment in the Improve phase of RDMAICS. Analyse has handed you a cause you believe in. The team has generated several ways to attack that cause. You cannot pilot all of them at once, and you should not try. The Pugh Matrix forces a comparison on the criteria that matter instead of a debate about who argued hardest.

The tool is deliberately coarse. You are not estimating savings to two decimal places. You are asking whether each option is better, the same, or worse than today on each criterion. That coarseness is the point. It keeps the conversation moving and keeps weak options from hiding behind spurious precision.

Use it after Analyse, before you build a pilot plan. Use it again if a pilot fails and you need to fall back to the next ranked option.

The example in this guide is the sample data built into the tool. It is fictional. No real organisation, product or team is described. It exists so you can see the mechanics end to end before you load your own numbers.

Picture a document handling process. Analyse has pointed at manual data entry and manual routing as the cause of errors and delay. Three concepts are on the table. The first is OCR Capture, which reads documents automatically instead of relying on typing. The second is Auto-Routing, which sends each document to the right queue without a person deciding. The third is Match Rules, which applies logic to pair documents with their records.

Five criteria were agreed before any scoring began. Error reduction. Cycle time reduction. Implementation cost. Time to deploy. Sustainability. Each one has a weight, and the weights are not equal.

The baseline column sits at zero across all five rows. That is by design. The baseline is the reference point, not a competitor.

The worksheet has three parts. You name the criteria. You weight them. You score each concept against the baseline.

In the sample data, error reduction carries a weight of 3. Cycle time reduction also carries a weight of 3. Implementation cost carries 2 and time to deploy carries 2. Sustainability carries 1. That spread says something plain about the project. Outcomes matter more than delivery effort, and delivery effort still matters more than nothing.

Then the scoring. OCR Capture earns a plus on error reduction, a plus on cycle time reduction, a minus on implementation cost, a minus on time to deploy, and a plus on sustainability. Auto-Routing earns a zero on error reduction, then a plus on cycle time reduction, implementation cost, time to deploy and sustainability. Match Rules earns a plus on error reduction, cycle time reduction, implementation cost and sustainability, and a zero on time to deploy.

Notice how much information a zero carries. Auto-Routing scores zero on error reduction because moving a document to the right queue faster does not change what a person typed into it. That is an honest answer, not a gap in the analysis. Resist the urge to turn every zero into a plus.

The tool recalculates from the editable worksheet inputs and produces a ranking. In the sample data, Match Rules sits at the top, Auto-Routing second, and OCR Capture third. The result chart runs on a scale marked at 0, 3, 6, 9 and 12.

Work out why the order fell that way. Match Rules collects pluses on four criteria, including both of the heavily weighted outcome criteria, and takes no minuses at all. Auto-Routing also takes no minuses, but it gives up the error reduction criterion with a zero, and that criterion carries a weight of 3. OCR Capture picks up pluses on both weighted outcome criteria, then loses ground with minuses on implementation cost and time to deploy.

So the ranking is not a verdict on technical merit. It is a verdict on this set of criteria at these weights. Change the weights and the order can move. That is a feature, because it makes your priorities visible and arguable rather than buried.

The tool also offers an AI review that explains the calculated result in context. Read it as a second pair of eyes on your reasoning, not as the decision.

A ranking tells you the order. It does not tell you the size of the prize. You still have to size the gain before committing resources, and the matrix gives you a useful starting frame.

Take the top ranked concept and read across its row of pluses. Match Rules claims improvement on error reduction and cycle time reduction. Those are the two criteria your Measure phase already instrumented. So you already know the current error rate and the current cycle time. Your pilot hypothesis writes itself. If Match Rules is a genuine plus on both, the measured baseline should move on both.

Now read the zeros and minuses as cost and risk. Match Rules carries a zero on time to deploy, which means deployment effort is no better than today. Budget for that. OCR Capture carries two minuses, which means if you ever pilot it, you need an explicit cost and schedule case before anyone signs off.

The discipline is simple. Pluses become the metrics you will track in the pilot. Zeros and minuses become the constraints you will plan around.

The first decision is what to pilot first. In the sample, that is Match Rules. It wins on the weighted criteria and carries no minuses, so the downside exposure is limited.

The second decision is sequencing. Auto-Routing ranks second and takes no minuses either. It also scores zero where Match Rules scores plus. Those two are not in conflict. They attack different parts of the process, which makes them candidates for a phased rollout rather than a straight choice.

The third decision is what to defer. OCR Capture ranks third because its minuses on implementation cost and time to deploy pull against its pluses. Deferring is not rejecting. You keep the row in the worksheet, and if cost or deployment conditions change, you rescore and recalculate.

The fourth decision is about the criteria themselves. If the ranking feels wrong to the room, the usual cause is a weight that does not reflect the real priority. Fix the weight and rerun. Do not quietly adjust a score to force the answer you wanted.

The most common mistake is scoring before the cause is confirmed. The Pugh Matrix compares solutions. It does not test whether those solutions address the real driver. If Analyse was thin, the matrix will rank the wrong family of options with confidence.

The second mistake is weight inflation. When every criterion carries a high weight, nothing is prioritised and the totals compress. The sample uses a 3, 2, 1 spread, with sustainability deliberately at 1. Somebody made a call, and the matrix records it.

The third mistake is avoiding the minus. Teams who like an idea will score a hard cost as a zero. OCR Capture takes minuses on implementation cost and time to deploy, and those minuses are exactly why it ranks third. Strip them out and the tool loses its value.

The fourth mistake is scoring concepts against each other instead of against the baseline. Every column must be judged against the same zero line. Mix the reference point and the arithmetic means nothing.

The fifth is treating the output as final. The worksheet is editable for a reason. New information arrives, you rescore, the ranking rebuilds.

The Pugh Matrix will not quantify a benefit. The result in the sample is a ranking across a scale marked 0 to 12. Those positions are weighted score totals, not dollars, not hours, not defect counts. If you need a financial case, build it separately using your measured baseline.

It will not validate your criteria. The tool takes the five criteria you typed and weights you set, and does the arithmetic faithfully. If you left out a criterion that matters, such as a regulatory constraint or a capability you do not have, the matrix cannot know and the ranking will look clean while being incomplete.

It will not account for interactions between concepts. Each column is scored in isolation. In reality, deploying Auto-Routing may change the cost or deployment time of Match Rules. The matrix has no way to represent that. You handle it in your implementation planning.

It will not tell you whether the top concept will work. A plus is a judgement that something should be better than today. The pilot is what turns that judgement into evidence. The matrix narrows the field so the pilot is worth running. Nothing more, and nothing less.

Finally, it will not remove the need for a Control phase. Ranking a concept and sustaining the gain are different problems. Take the top ranked option into a pilot, measure against the baseline you already captured, and let the data decide what you standardise.