A customer buys a little more than usual for two weeks.
Purchasing increases the next order to be on the safe side.
The supplier sees an order twice the size of the last one and ramps up production.
Three months later, you and your supplier are both sitting on goods nobody needs.
That is the bullwhip effect in supply chains.
Small changes in customer demand can grow into large swings in purchasing and production. The result can be excess inventory, delays and unnecessary costs.
Here, we look at why it happens and what you can do to stop your own planning from amplifying the swings.
The bullwhip effect describes how orders placed with a supplier vary more than sales to customers, and how that variation grows the further upstream you go in the supply chain.
The name comes from a whip: a small flick of the wrist creates a much larger movement at the other end.
Jay Forrester described the phenomenon in his 1961 book Industrial Dynamics, which is why it is also called the Forrester effect. The term "bullwhip effect" came later, partly through research at Stanford.
There is an important distinction here. Demand variability is the variation in what customers actually buy. You cannot eliminate it completely.
The bullwhip effect is the extra variation that arises when each link in the chain reacts to orders from its immediate customers instead of to actual demand. That is the part you can influence.
In 1997, Hau Lee, V. Padmanabhan and Seungjin Whang showed that the bullwhip effect is not just the result of poor decisions. Each link can act rationally based on what it sees and still amplify the variation. They identified four causes:
Long and variable lead times make all of this worse.
When lead times are uncertain, replenishment becomes harder to plan. Companies may respond by increasing their orders or inventory buffers, which can amplify the swings.
Consider a wholesaler that sells around 100 units a week of a particular product. The wholesaler normally orders in batches of 300 units.
In weeks 6 and 7, the sales team runs a promotion and sales increase. Purchasing reacts by placing an order for 600 units in week 7, twice the normal order size, to be on the safe side.
| Week | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Sales to customers (units) | 98 | 104 | 97 | 101 | 99 | 150 | 135 | 92 | 88 | 95 | 100 | 101 |
| Orders placed with supplier (units) | 0 | 300 | 0 | 0 | 300 | 0 | 600 | 0 | 0 | 0 | 0 | 300 |
Over the 12 weeks, the wholesaler sells 1,260 units but orders 1,500.
If all deliveries are received within the period and there are no other inventory movements, inventory has risen by 240 units. Whether that constitutes excess inventory depends on future demand. But the example shows how one extra order can increase inventory levels, even though sales return to their previous level after the promotion.
Sales to customers vary relatively little, while orders to the supplier vary sharply. Measured by relative variation, orders vary roughly nine times more than sales.
But there is an important distinction. Much of the variation comes from the wholesaler ordering in fixed batches of 300 units. That does not necessarily have to be a problem if the supplier knows the ordering pattern and can plan around it.
The ratio tells you that variation exists. It does not tell you what is costing you money.
You can measure the bullwhip effect by comparing how much orders placed with the supplier vary relative to sales to customers. One way to do this is with the coefficient of variation, which shows how much a set of values varies relative to its mean.
You calculate the ratio by dividing the coefficient of variation for orders by the coefficient of variation for sales. That is also the method used in the example above.
A result above 1 means that orders vary more than sales in relative terms.
In Excel, you calculate the coefficient of variation for a series of weekly values with =STDEV.S(B2:B13)/AVERAGE(B2:B13). Calculate it for both sales and orders over the same period, then divide the result for orders by the result for sales.
A few things can distort the picture. If you measure weekly and order in batches, batching alone can produce a high ratio even when nothing is wrong. If the product is seasonal, the period should be long enough to show the pattern. And because the ratio cannot distinguish planned batches from reactive orders, always look at the orders themselves afterwards. When were the large orders placed, and what happened just before?
The bullwhip effect does not appear as a separate line in your financial statements. It hides in other numbers.
You see it as excess inventory after promotions and as rush orders when the next delivery arrives too late. You see it in safety stock that keeps growing because nobody trusts the forecast. And you see it in your supplier's response: longer lead times, higher minimum orders or lower OTIF because your order pattern is hard to plan around.
Instability breeds more instability. Shortages lead to bigger orders. Bigger orders lead to excess inventory. Excess inventory leads to a period with no orders, and then the cycle starts again.
You will not eliminate it completely. Customer orders will always vary, and some batching makes sense. But you can stop amplifying the swings. Start where your own data show the signal getting distorted.
Base your forecast on actual sales to customers, not on the orders coming in from resellers or internal warehouses. If you are a wholesaler or manufacturer and can get sales data from your customers, those figures often give you a clearer picture of demand than their orders alone.. Demand planning is more effective when the forecast separates trend, seasonality, promotional activity and random noise.
A promotion can create extra sales or bring demand forward from a later period. If the supply chain team does not know that weeks 6 and 7 were promotion weeks, the increase in sales may be mistaken for sustained growth. Flag promotional periods in your data and discuss them as part of your S&OP process, so purchasing understands the context before placing orders.
The obvious fix is smaller, more frequent orders. That can reduce order variability, but it also means more orders, more freight and more handling, and it can cost more than it saves.
As the example showed, a fixed batch size is not necessarily a problem if the supplier knows the ordering pattern and can plan around it. Share your expected ordering schedule with the supplier and avoid changing it in response to a single week's sales fluctuation.
Your supplier may only see the orders you place. If you share forecasts, promotion plans and inventory status, the supplier can better distinguish a sustained increase in demand from a one-off order. One way to take this further is through vendor managed inventory (VMI), where the supplier takes responsibility for replenishment.
Safety stock should protect against genuine uncertainty in demand and lead time. It should not compensate for uncertainty you create yourself by overreacting. If your supplier's lead time varies significantly, investigate whether that variation can be reduced before you automatically increase safety stock. Also bear in mind that just-in-time with very small buffers can make you more vulnerable to disruptions further upstream in the supply chain.
The calculation for one SKU takes ten minutes in Excel. The hard part is doing it for thousands of SKUs, fidentifying where order variability exceeds sales variability, and seeing whether the cause is a promotion, order batching or supplier performance.
Once you can distinguish changes in actual demand from the variation your own decisions create, you can start addressing the causes rather than just reacting to the numbers.