Works with Claude, ChatGPT, Gemini, or your own local model: paste this prompt and your AI will co-facilitate this exercise with you.

Help me prepare and run the exercise below. First ask for my participant count, time, setting, access needs and intended result. Check these against the stated limits. Use the supplied rules and materials. Explain any proposed changes before using them, and label them as adaptations. Do not invent source claims or require personal disclosure. Give me one step at a time when I say start. I manage the people, physical activity and clock; do not claim to observe the room. Ask me what happened before choosing a next step.

Reference revision: 040f33dcf6b5e9b18c3614e3fc783559c8dfb565c250247016d930815ccc65e6

# Count the Beans

Participants estimate the number of beans in a clear jar alone, negotiate shared estimates in expanding groups, and compare their collective numbers against the true total.

**Time:** 15 minutes. UWT suggested 15-minute pacing to allow individual, pair, small group, and whole-room debrief.
**People:** 4 to 24. Individual work, then pairs or trios, then combined small groups of roughly 4 to 6, then whole room.

## Choose this exercise

**Purpose:** Examine how negotiation, aggregation, and group size affect estimation accuracy and consensus.
**Use it when:** Use when a group needs to examine decision-making dynamics, the wisdom of crowds, or how voices negotiate numbers under uncertainty.
**Skip it when:** Skip when the group is working on a technical problem with strict algorithmic solutions or when participants lack time to engage in multi-stage group discussion.

## Materials and tools

- 1 transparent jar filled with a known count of dry beans, marbles, or hard candy
- 1 piece of scratch paper per participant
- 1 pen or pencil per participant
- 1 shared whiteboard or flip chart and marker to record group estimates
- 1 index card to conceal the true count

## Before participants arrive

1. Count the items into the jar beforehand and seal the lid securely.
2. Record the verified total on an index card hidden from participant view.
3. Place the jar on a central table visible to all participants.

## Say this to open

Take a look at this jar. Without touching or opening it, you will guess how many beans are inside. We will start with individual estimates, merge those into pair estimates, combine into larger groups, and see how our numbers shift as groups grow.

## How to run it

### 1. Individual estimation (2 minutes)

Display the jar where everyone can view it. Direct each participant to write down their private estimate on scratch paper without talking. State that passing on guessing is fine, and anyone choosing an observer or scribe role is not required to record a private number.

### 2. Pair or trio consensus (3 minutes)

Ask participants to turn to a partner. If there is an odd total, form one group of three. Have each group compare their numbers and agree on a single consensus estimate. For example, if one guesses 300 and one guesses 500, they might average to 400 or debate container volume. They write this single agreed number down.

### 3. Small group consensus (4 minutes)

Combine adjacent pairs or trios into groups of roughly 4 to 6 people. If merging leaves a leftover pair or single trio, fold them into existing groups even if a group reaches 7, or leave one trio if the total count is very small. Each combined group negotiates a single group estimate. Have each group designate one person to write their agreed number on the whiteboard.

### 4. Whole-room consensus and reveal (3 minutes)

Look at the small group estimates on the board. Give the room two minutes to agree on one single final estimate for the entire room, or take the median of the group estimates if time runs short. Reveal the true count from your hidden index card.

### 5. Debrief (3 minutes)

Lead a brief reflection on how numbers changed as groups expanded. Take two or three responses from the room based on the debrief questions.


## Debrief questions

- How did your estimate change as you moved from individual to pair to small group?
- What method did your group use to reach an agreed number: mathematical averaging, persuasion, or compromise?
- Did the larger group estimate get closer to or farther from the real total than your individual guess?

**Finished output:** A series of progressive consensus estimates recorded on a whiteboard compared against the actual verified count.

## Remote

Hold the jar up to the webcam at multiple angles or provide clear photographs showing top, side, and scale (such as a standard pen next to the jar). Participants post individual estimates via direct message to the facilitator, join breakout rooms of two or three, return to join breakout rooms of four to six, and post group estimates into the main chat.

## Hybrid

In-person participants inspect the physical jar while remote participants view a high-resolution live video feed and still photos with a scale reference. Remote participants send individual estimates via direct message to the facilitator. Pair remote participants with remote participants in breakout rooms and in-person participants with each other to keep audio simple. Both groups post small group numbers to a shared online whiteboard.

## Access and participation choices

- Provide a clear close-up view or pass the jar around so people with limited distance vision can inspect it.
- Anyone may choose to observe the group debate or act as the team scribe rather than offering a personal numeric guess. Those taking an observer or scribe role do not need to record a private estimate.

## If the session gets stuck

**A group cannot reach consensus and starts arguing over mathematical precision.**

Remind them that time is limited and suggest taking the mathematical mean or median of their numbers so they can proceed.

**One dominant voice forces their original estimate onto the rest of the small group.**

During the debrief, ask groups how power dynamics and confident voices influenced the negotiated numbers compared to simple pooling.


## Terms used here

- Consensus estimate: A single agreed-upon number produced by discussion and mutual agreement, rather than a forced vote.

## Sources and adaptation

Observed on Playmeo, where the activity is attributed to Mark Collard.

Retained the core Playmeo progressive grouping mechanic (individual, pairs, expanding groups, whole group consensus). UWT added explicit subgroup merging logic to handle uneven participant counts, clarified observer roles to eliminate conflicting guess requirements, added remote and hybrid procedures, and structured time-boxed steps.

[Playmeo](https://www.playmeo.com/activities/team-building-problem-solving-activities/count-the-beans)

[Playmeo](https://www.playmeo.com/activities/team-building-problem-solving-activities/count-the-beans/)

Current run sheet: https://unitedwetransform.com/exercises/count-the-beans-team-based-variation-of-classic-guessing-game/