How I handle duplicate ideas in workshops

A workshop can produce hundreds of cards. That does not mean it has produced hundreds of different ideas. Several people may have suggested the same action, each in their own words. Others may have added a useful detail to a similar suggestion.

I want to keep those contributions. I also need to make the content easy to discuss and prepare clear choices before a vote. This is why we added card stacks to Stormz.

Repeated ideas make discussion and voting harder

Take two answers to a question about improving meetings:

• “Send the agenda before the meeting.”
• “Share the meeting agenda in advance.”

They express essentially the same action. On a board with many cards, I have to notice that connection while also reading the other contributions and following the conversation.

During a debrief, presenting each card separately can mean hearing the same point several times. The useful differences become harder to notice: someone may have explained why the agenda matters, or given an example of what happens without one.

Then comes the vote. If equivalent choices remain on separate cards, participants may divide their votes between them. Support for one action is spread across several places, which can make another action look more popular at first glance.

I have had this problem in workshops. I do not want a proposal to appear weaker simply because several people contributed it independently.

This matters in person and remotely. A large plenary session can produce many similar contributions. So can small teams brainstorming separately and bringing hundreds of ideas back to the same board.

A stack keeps the original contributions

One way to deal with duplicates is to merge them into a new card. But then I have to write a version that represents everyone. A detail can disappear. A meaningful difference can get lost in the new wording.

Another option is to choose one card and set the others aside. That makes the main board easier to read, but the contributions I set aside become less visible.

A stack gives me a different option. One original card appears on top, with a count showing that there are more cards underneath. All the cards remain accessible. I can change the top card, remove a card from the stack, or separate the whole stack later.

Each card keeps its own text, author, tags, categories, comments, and votes. Nothing needs to be rewritten to make the stack. The top card does not inherit everyone else's votes or combine all their tags.

That is the design choice I like: I can show fewer cards on the board without throwing away the information behind them.

I prepare the stacks before the vote

My aim is to make each choice easy to understand. I look for cards proposing the same action, put them in a stack, and choose the clearest card to show on top. Then I check that the other cards really belong there.

“Share the agenda in advance” and “Make meetings shorter” both concern meetings. They are still different actions. Before a vote, I want people to be able to choose between them, so I keep them separate.

I also use stacks for closely similar or complementary ideas during a discussion. I am more careful with those before voting. A useful stack for a conversation does not always represent one choice.

Related cards helps with the debrief itself. While someone presents their card, I can see who else has written something similar and invite them directly:

“I see Lisa added something similar. Lisa, would you like to tell us about your idea?”

I can see whom to invite next without searching the whole board. The conversation can then bring out a detail or a difference that a quick reading might miss.

Before starting the vote, I review the top cards as the choices people will see. Stacking does not combine votes already cast. The benefit comes from preparing one visible choice for each repeated idea, rather than asking people to vote across several cards for the same action.

AI saves time, and I check the stacks during a break

When there are hundreds of cards and only a few minutes before a vote, doing all of this manually can be difficult. AI stacking gives me a way to prepare stacks quickly.

The trade-off is that I am asking AI to decide which contributions belong together. It can put cards in the same stack because they sound similar, even when they suggest different actions. It can also miss a connection that is obvious to someone who heard the discussion.

My usual approach is to start with AI, then take advantage of a break to fine-tune the stacks manually. I read the cards inside each stack, separate choices that should remain distinct, and check which card appears on top.

Because stacks preserve the originals, correcting an AI decision is straightforward. I do not have to reconstruct contributions from a merged summary.

In a time-limited workshop, that combination matters. I can prepare the board in a few minutes and spend my attention on the decisions that need checking. I still want a human review before asking participants to vote.

Stacking and clustering serve different purposes

I use clustering to create meaningful categories that help the workshop move forward. A cluster is a bucket: the participants decide which cards belong together and agree on a name. Those names influence how we interpret the content and what we discuss next.

For example, “We communicate but we don’t listen” names a pattern that a label like “Communication” leaves unexplained. It gives us something to discuss: where are we failing to listen, and what needs to change?

The participants choose that label together. They discuss what the contributions mean and agree on the pattern they see. My role is to help that discussion.

A stack still appears as a card. It does not create another category system. The cards keep their existing categories and tags, and the top card appears on the board with its own information.

This lets me reduce repetition while keeping the questions, categories, and tags I already use to organize the activity.

It also explains why I am more inclined to use AI for stacking than for clustering. Checking whether two cards propose the same action is a narrower task than deciding which categories will help people understand a complex situation. Clustering needs the participants to make sense of the ideas together and agree on what they mean.

Both tools have a place. I use stacks to keep repeated or closely similar contributions together. We use clusters to decide together which larger patterns will help us move forward. Stormz lets me do both without having to sacrifice the original cards.