Nominal Group Technique: Turn a Diagnosed Cause Into a Change Worth Piloting

Nominal Group Technique uses structured voting to rank solution ideas by team consensus. It balances individual input with group prioritisation. Each participant scores every idea alone. The tool then totals the scores and ranks the list.

You reach for it in the Improve phase of RDMAICS. By then you have a diagnosed cause from Analyse. You also have more ideas than capacity. The question is no longer what is wrong. The question is which change you pilot first.

The structure matters. Silent, individual scoring comes before the group sees any total. That order limits the pull of the loudest voice in the room. It also gives you a record of where the group agreed and where it did not.

The screenshots in this guide show the tool loaded with its built-in sample data. The case is fictional. Treat it as a worked example, not as evidence from a real site.

The scenario is an invoice processing workflow. The team has finished Analyse and landed on manual handling as the diagnosed cause. Six solution ideas are on the table. Three participants score each one.

The ideas span capture, routing, matching, vendor data and approvals. OCR-based invoice capture. Auto-routing with delegation. Automated 3-way match rules. Vendor portal plus validation. Mobile approval app. Vendor master cleanup. Each idea is plausible. None of them can all be piloted at once.

Start by listing the solution ideas, one per row. Keep each one specific enough to pilot. An idea you cannot describe in a single line is usually two ideas.

Add a column for each participant. The sample uses three. Every participant scores every idea on the same scale. In the sample data the scores run from 1 to 5. The highest individual score given is 5. The lowest is 1.

Then let each participant fill their column without seeing the others. This is the part people skip. If participants score together, the technique collapses into a show of hands. The totals will look the same. The thinking behind them will not.

The tool recalculates the totals and the ranking from the editable worksheet inputs. You can change a score and watch the order move.

The tool totals each row and ranks the ideas by total score. In the sample data the order comes out like this.

OCR-based invoice capture scores 5, 5 and 4 for a total of 14 and rank 1. Auto-routing with delegation scores 4, 5 and 4 for a total of 13 and rank 2. Automated 3-way match rules scores 3, 4 and 4 for a total of 11 and rank 3. Vendor portal plus validation scores 3, 3 and 3 for a total of 9 and rank 4. Vendor master cleanup scores 2, 1 and 3 for a total of 6 and rank 5. Mobile approval app scores 1, 2 and 2 for a total of 5 and rank 6.

The chart plots the totals from highest to lowest across a scale running from 0 to 16. The gap between the top of the list and the bottom is visible at a glance. The tool also offers an AI review that explains the result.

Look at the shape, not only the order. The top two totals, 14 and 13, sit one point apart. The bottom two, 6 and 5, also sit one point apart. The middle of the list is where the real separation happens, between 13 and 11 and then between 9 and 6.

A rank is not a business case. It is a shortlist. The sample result tells you the group places OCR-based invoice capture and auto-routing with delegation above the rest. It does not tell you what either one is worth.

So size the gain before you commit. Take the top ranked idea and estimate what it changes in the measure you baselined during Measure. Count the volume it touches. Estimate the handling time it removes. Put a range around the estimate rather than a single figure.

Then check whether the second ranked idea would be cheaper to test. With a total of 13 against 14, auto-routing with delegation is barely behind. If it pilots in a fraction of the time, it may be the better first move. Consensus ranking and pilot sequencing are two different decisions.

Finally, write down what the pilot has to achieve to be judged a success. Do that before the pilot starts. A target set afterwards is not a target.

The ranked list drives four practical calls. The first is what to pilot now. In the sample that is the idea at rank 1 with a total of 14.

The second is what to park with a reason. Mobile approval app sits at rank 6 with a total of 5. Parking it is defensible because three people scored it independently and none of them placed it high. That record saves the conversation from being reopened every fortnight.

The third is where to look harder. Vendor master cleanup carries scores of 2, 1 and 3. The spread is wider than anywhere else in the sample. One participant sees something the others do not. That is worth five minutes of discussion before the list is closed.

The fourth is how to sequence. Ranks 1 to 3 total 14, 13 and 11. They cluster. If capacity allows a second pilot, you take it from that cluster rather than from the bottom of the list.

The most common mistake is scoring in public. Participants anchor on the first number they hear. The independent column is the whole point of the technique.

The second is voting on ideas that do not address the diagnosed cause. Nominal Group Technique ranks what you put in front of it. If an unrelated idea enters the list, it can still finish high. Filter the list against the Analyse output first.

The third is treating the total as precision. A total of 14 and a total of 13 are close. Reading a one point difference as a clear winner overstates what three scores can carry.

The fourth is a vague idea. Vendor portal plus validation, at a total of 9, could mean several different builds. If participants score different mental pictures, the total means little.

Nominal Group Technique measures group preference. It does not measure effect. Nothing in a total score of 14 proves that OCR-based invoice capture will reduce cycle time. The ranking reflects what three people believed when they scored.

It does not price anything. There is no cost, no effort estimate and no payback in the output. You add those outside the tool and you add them before you commit resources.

It does not fix a weak idea pool. If the six options on the worksheet are all mediocre, the ranking will still produce a rank 1. The tool orders what it is given.

It does not account for dependencies. In the sample, vendor master cleanup sits at rank 5 with a total of 6. If clean vendor data is a prerequisite for a higher ranked idea, the sequence has to override the rank. The tool will not flag that for you.

Use it for what it is good at. It converts scattered opinion into an ordered, recorded shortlist, quickly and without the loudest voice deciding. Then do the sizing work yourself.