The Beer Game

The Beer Game is a supply chain simulation created at the MIT Sloan School of Management in the 1960s, associated with Jay Forrester. Participants take on four successive roles: retailer, wholesaler, distributor, and factory. Each simulated week, they receive deliveries, serve their customer, and place orders with their supplier, without direct communication between roles. The game almost systematically reveals stockouts, overstocking, and amplified oscillations up the chain. It demonstrates that the structure of the system, delays, and lack of information often produce observed behaviours more than individual intentions.

Walkthrough

1. 1

   Frame the experience and set up the chain

   10 to 15 min

   The facilitator presents the challenge: "You will manage a beer supply chain, week after week, with a single objective: to minimise the total cost of the chain." They form chains of four roles: retailer, wholesaler, distributor, and factory; with more than four participants, multiple chains play in parallel or pairs take on one role. They clarify that the end customer orders from the retailer, and then each role orders from its upstream supplier. Participants arrange themselves in the order of the chain to visualise the flows clearly.

   Tip — Physically represent the direction of flows with two visible arrows: orders upstream, deliveries downstream. This reduces errors without providing additional information.

2. 2

   Explain the operational rules

   15 to 20 min

   The facilitator states the rules without interpreting them: "Each round represents a week. Each week, your role receives the expected deliveries, fulfills customer orders as much as possible, notes any late orders if stock is insufficient, and then places an order with your supplier." Orders and deliveries are subject to delays of several weeks, as specified by the materials or platform used. Roles do not communicate with each other: no discussion about stocks, intentions, orders, or forecasts. The only costs to track are held stock and late orders, with the latter costing twice as much as stock.

   Tip — Ask a participant to rephrase: "What do you do in a game week, in what order?" Correct the order immediately, as an inversion can quickly distort the simulation.

3. 3

   Launch the first stable weeks

   15 to 25 min

   The facilitator starts the first rounds with a stable final customer demand, according to the chosen scenario. They guide the pace: "Receive your deliveries, serve your customer, note any delays, then write your supplier order." Participants are not yet trying to optimise finely; they are learning the mechanics and observing their stock levels. The facilitator only checks for adherence to the silence between roles and the proper maintenance of tracking sheets.

   Tip — At the beginning, strongly pace each sub-action out loud. Once the group is fluid, slow down your interventions to allow spontaneous decisions to emerge.

4. 4

   Introduce the one-time demand increase

   5 min

   At the week specified by the scenario, the facilitator applies the one-time increase in final customer demand. They do not announce it to the entire chain as a strategic event; it simply enters through the retailer, who only sees the customer order. They remind: "The demand increases only once, then remains stable. The other rules do not change." Other roles continue to receive only the orders from their direct customer, with transmission and delivery delays.

   Tip — Maintain a neutral tone at the time of the increase. If you dramatise, you influence the orders and weaken the educational power of the outcome.

5. 5

   Continue until stockouts and overstocking

   30 to 60 min

   The facilitator continues the weeks at the same pace, without allowing communication between roles. Participants often observe growing orders, accumulating delays, and then massive arrivals that create overstocking. The facilitator does not advise on strategies: they simply enforce the rules, timelines, and order of operations. They invite everyone to continue making local decisions: "Do what seems right with the information you have."

   Tip — When tension rises, protect the simulation: humorously but firmly prohibit side conversations. The frustration of lack of information is a crucial aspect of the game.

6. 6

   Calculate costs and visualise the result

   10 to 15 min

   At the end of the last week, each role totals its costs of stock and late orders. The facilitator reminds the scoring system: each unit in stock has a weekly cost, and each unit of late order costs twice as much; the total cost of the chain is the sum of the costs of the four roles. Teams compare their results if multiple chains have played. The facilitator displays or charts the evolution of stocks, delays, and orders to show the amplification as it moves up to the factory.

   Tip — Do not limit yourself to the final score. Have participants look at the curves by role: this is often where the bullwhip effect becomes visible and indisputable.

7. 7

   Debrief the structure of the system

   20 to 30 min

   The facilitator opens the debrief by starting from the experience: "What did you try to do, and what actually happened?" They bring out the stockouts, overstocking, and increasingly strong oscillations up the chain. They then refocus the learning: most players make reasonable local decisions, but the structure of the system produces the overall behaviour. They conclude on the combined roles of delays, lack of information, and local optimisation.

   Tip — Avoid judgments like "you ordered too much." Replace them with "what information did you have at that moment?" to shift the analysis from individuals to the system.

Variants

• Have multiple chains play in parallel, then compare total costs and curves. The debrief shows that similar behaviours emerge even with different teams.
• Conduct a short second round after the debrief, with a shared information rule chosen by the group. The comparison allows for a concrete test of the impact of visibility on the system.
• In a remote setting, use a prepared platform or shared spreadsheet with the four roles, delay queues, stocks, delays, and orders. Place each role in a separate channel and prohibit cross-messaging during the simulation.
• In a large group, assign each role to a pair: one person decides, the other keeps accounts and checks the order of operations. This increases reliability without removing local responsibility.