Multi-Voting: Turn a Long Solution List Into the Few Worth Piloting

By the time you reach the Improve phase of RDMAICS, you have a diagnosed cause. You also have a problem of abundance. A workshop with eight people will produce twenty solution ideas inside an hour. Most of them are plausible. Few of them are fundable.

Multi-Voting exists to close that gap. The tool reduces a long solution list to a prioritised few through successive rounds of dot voting. Each participant gets a fixed number of votes. They spend them on the ideas they believe are worth pursuing. The lowest scoring ideas drop away. The survivors go to another round with fewer votes available, which forces harder choices.

The value is not the arithmetic. The value is the conversation that happens between rounds. When an idea loses half its support from one round to the next, somebody in the room has changed their mind, and the reason why is usually worth hearing. Multi-Voting gives that reason a place to surface before money is committed.

Use it when you have more candidate solutions than you have capacity to pilot. Do not use it to choose between two options. Two options need a direct comparison, not a vote.

Improve sits after Analyse for a reason. You should arrive with evidence about what is causing the problem. Multi-Voting assumes that evidence exists. It prioritises responses to a known cause. It does not help you find the cause.

The tool works best at a specific moment. You have run the cause analysis. You have held a solution generation session. The list is long, the enthusiasm is high, and nobody wants to be the person who says no. Multi-Voting makes the no collective rather than personal.

It also works as a sizing instrument. Vote counts tell you how much conviction sits behind each idea. An idea that holds its votes across three rounds carries broad support. An idea that collapses carries a single champion. Both are useful signals when you decide where to spend pilot effort.

The tool ships with sample data built around invoice processing. This example is fictional. It is supplied to show the mechanics of the tool, not to report a real result from a real organisation.

The scenario runs like this. A finance team has a backlog in accounts payable. The Analyse phase has pointed at manual handling and approval delay. A solution session has produced six candidate ideas. Each one is a credible response to the diagnosed cause, and each one would consume real build effort.

The six ideas on the list are OCR-based invoice capture, auto-routing with delegation, automated 3-way match rules, vendor portal and validation, a mobile approval app, and vendor master cleanup. Three rounds of voting follow. Nothing is eliminated by decree. The votes do the narrowing.

The worksheet is deliberately plain. Each row holds one solution idea. Each column holds one round of votes. A final column records whether the idea advances.

In the sample, round one spreads votes across all six ideas. OCR-based invoice capture takes 9. Auto-routing with delegation takes 8. Automated 3-way match rules takes 6. Vendor portal and validation takes 5. The mobile approval app takes 3 and vendor master cleanup takes 2.

Round two narrows the field. The two lowest scoring ideas fall to 0. The remaining four ideas each lose a vote compared with round one, which is what you expect when the vote allowance shrinks. OCR-based invoice capture holds 8. Auto-routing with delegation holds 7. Automated 3-way match rules holds 5. Vendor portal and validation holds 4.

The final round confirms rather than reshuffles. The final votes match round two exactly: 8, 7, 5, 4, 0 and 0. Stability across the last two rounds is a meaningful outcome in itself. It tells you the room has settled.

The result view ranks the ideas by final votes and recalculates from the editable worksheet inputs. The chart runs on a scale from 0 to 8, and the bars fall in a clear order.

Three ideas are marked to advance. OCR-based invoice capture at 8, auto-routing with delegation at 7, and automated 3-way match rules at 5. Three are marked not to advance. Vendor portal and validation at 4, the mobile approval app at 0, and vendor master cleanup at 0.

The interesting line is between 5 and 4. Automated 3-way match rules advances with 5 final votes. Vendor portal and validation does not advance with 4. One vote separates them. That is not a mandate. It is a prompt to discuss, and the tool makes the closeness visible rather than hiding it inside a cut-off.

The tool also offers an AI review that explains the ranked result. Treat it as a second reading of the same numbers, not as a decision.

Multi-Voting supports a small set of decisions well. The first is scope. Three ideas advance out of six, so your pilot plan covers half the list rather than all of it. That is a concrete reduction in commitment.

The second is sequencing. The ranked order gives you a default sequence. OCR-based invoice capture leads with 8 final votes, so it is the natural first pilot. Auto-routing with delegation follows at 7. Automated 3-way match rules sits third at 5.

The third is the parked list. Ideas that fall to 0, such as the mobile approval app and vendor master cleanup, are not dead. They are recorded with their round one scores of 3 and 2. If the first pilot changes the picture, you can revisit them with that history intact.

The fourth is the borderline conversation. Vendor portal and validation sits at 4 and does not advance. You should decide explicitly whether to park it or to run it as a smaller trial. The tool surfaces the question. You answer it.

The most frequent error is voting before the ideas are understood. If half the room does not know what automated 3-way match rules means in practice, the votes measure familiarity rather than merit. Explain every item on the list before any dot is placed.

The second error is skipping the discussion between rounds. The drop from round one to round two is where the reasoning lives. In the sample, two ideas go from 3 and 2 down to 0. Somebody should say out loud why.

The third error is treating the vote total as a business case. A score of 8 is a measure of support in one room on one day. It is not a measure of benefit, cost or feasibility. Multi-Voting narrows the field so that the proper sizing work can be done on three ideas instead of six.

The fourth error is loading the room. If the people who proposed the ideas outnumber the people who do the work, the votes will reflect authorship rather than impact.

Multi-Voting will not tell you whether an idea works. It records opinion, converted into a count. The sample shows OCR-based invoice capture leading at 8 final votes. That number says the room believes in it. It says nothing about whether capture accuracy will hold on real invoices.

The tool will not size the gain. There is no benefit figure, no cost figure and no time estimate anywhere in the worksheet. Sizing the gain happens after the vote, on the shortlist, using whatever measurement you established in the Measure phase. Multi-Voting reduces the number of ideas you need to size. It does not do the sizing.

It will not resolve close calls. The one vote gap between 5 and 4 in the sample is a tie in everything but arithmetic. The tool reports the gap. A human decides what to do about it.

It will not protect you from a poorly composed room. Votes from the wrong group produce a confident ranking of the wrong ideas. Nothing in the chart will warn you. Check who is voting before you trust what they voted for.

Finally, it will not replace the Control and Sustain phases. A pilot that wins the vote still needs a measurement plan and a handover. Multi-Voting gets you to the starting line of the pilot. The rest of RDMAICS gets you past it.