Advantages and Pitfalls of Push and Pull Strategies in Distribution Networks

addtoany linkedin

What’s driving your supply chain – immediate consumer demand or future projections? In either case, the goal is likely the same: to provide the best customer experience. A truly customer-oriented supply chain strives to fulfill the customers’ demands on-time. Success is defined by the on-time-delivery to request (OTD-R). In other words, when the product actually gets to the end consumer.

Supply chains are planned based on when a product is produced, delivered to distribution centers and made available at retail stores. The most common strategies for moving from upstream to downstream sites are push and pull strategies, or some mix of both. A pull strategy is when customer demand drives the entire production process. On the other hand, a push strategy is when production is based on long term customer forecasts.i

So which one is better? The answer is, it depends. There are pros and cons to using push vs. pull strategies within your distribution network. Let’s put the concepts of push and pull strategies into a real-world example. Assume you own a restaurant and for simplicity sake, you only serve breakfast. The main processes (Figure 1) to get a customer his or her order would be procurement (buying ingredients), manufacturing (cooking), and delivery (serving the customers).

Let’s say buying the ingredients takes an hour, cooking takes 20 minutes, and delivery takes another five minutes. Using a pull strategy in your supply chain, you wouldn’t start buying the ingredients until a customer had sat down at a table and ordered something. The result would be the customer receiving a fresh, high quality meal, and you carrying minimum inventory and having the flexibility to easily manage your processes.

But how many people do you know who are willing to wait nearly an hour and a half just to get their breakfast? That’s how long the entire process would take to go from procurement, to manufacturing, to delivery. Now think about what would happen if 10 customers showed up in intervals of 10 minutes apart and all ordered the same thing. You’d have to repeat the entire process from start to finish 10 times! That doesn’t seem very effective.

While there are many advantages to the pull approach – higher service levels, lower carrying costs, decreased inventory levels, and fewer markdowns – there are some serious drawbacks. A pull strategy starts to break down when lead times get longer and demand changes rapidly. It’s also a lot more difficult to take advantage of economies of scale because production and distribution are based on real demand, and therefore only scheduled as needed.ii

In our simple example, the time it takes to purchase the ingredients was the same as the number of customers – 10 customers, 10 hours spent buying ingredients. Push strategies on the other hand are driven by projections of customer demand. Using our same breakfast example, you’d prepare everything in advance based on the number of customers you expected to order each item.

You’d have ample time to purchase your ingredients (raw materials) and pre-cook the orders. When your customer arrives, it would only take five minutes to serve him or her breakfast. But what if your customer doesn’t show up? What if they want something different than you expected? What if they want to make a change to the item – adding extra cheese or making sure there are no onions? The answer is easy: There would be a lot of waste.

In push approaches, there’s a big emphasis on forecast. But a forecast is simply a guess because consumer-buying behaviors are not always predictable. In order for push planning to be successful, the forecast must be accurate. That alone can present a major challenge for today’s companies. Then you have added costs to store all the inventory, and as is the case with our example, the risk of goods expiring or spoiling before they’re sold. While not a problem for every industry, it certainly presents a very real challenge for those working in packaged foods, beverages, and even pharmaceuticals where expiration dates are prevalent.

The inventory carrying costs in a push supply chain are typically higher than those using a pull strategy. While our breakfast example certainly highlights the pitfalls associated with using a push or pull supply chain strategy, the reality is, most businesses are using a hybrid approach. They use a push-based system to get the products to regional warehouses or retailers, but from there they have to wait for a customer to "pull" a product.

Companies typically stockpile the finished product at its distribution centers to wait for orders that pull them in near customer stores.iii The rule of thumb is to stop pushing beyond the distribution center, but there are still some occasions to break this rule and push parts to retail stores. For example, for products close to their minimum shelf life, the distribution planner could trigger the push and pass them to a location close to the end customer. Or the distribution planner could take advantage of underutilized transportation and push the parts to downstream sites. But to do that successfully and without hours upon hours of manual calculations and analysis, having a supply chain software solution like Kinaxis RapidResponse with the ability to manage distribution requirements planning (DRP) is a necessity.

The ideal DRP application would recommend parts to be pushed past the distribution centers based on their expiry, storage utilization and any seasonality factor. When a part is expired within the distribution horizon, it would automatically be flagged by the system as a push candidate. Since storage utilization is another factor for pushing a part, when a part occupies more than a threshold percentage of storage it would also be flagged. And finally, if there is a seasonality impact within the distribution horizon, the part would be considered as a push candidate. Under these circumstances there’s no harm to push the parts to a location closer to end customer, and in fact could result in more positive business outcomes. What other factors might set a part as push candidate? Let us know in the comments!

i ii iii


Monti Lawrence
- September 14, 2016 at 9:56am
Great post! Thanks for the insights.
Peter Vollebregt
- September 15, 2016 at 10:11am
Nice basic write-up. But difficult for a lot of Supply Chains (IT systems) to dynamically change a part from Pull to a Push strategy. The one I know best (Oracle E-Business Suite) surely cannot handle it dynamically. Note that I think dynamic changes are important in the retail domain mainly (lot of different shops).

Regarding the last question about Push Candidates. I think service parts are typical Push candidates as they need to be there but demand is unpredictable. A kind of 2-bin system then sometimes works very well.
Krishna Madaiah
- September 18, 2016 at 6:03pm
Good Article,
The chain should consist of push and pull, As demonstrated by the example the ingredients should be push based on forecast or average consumption per day. However, the preparation of breakfast should be pull based, i.e only when there is an actual customer order. There is no point in preparing breakfast and reducing the shelf life of the product or not meeting the customer requirement.
The industry I work nervousness of the products has created the huge surplus of raw materials. As procurement looking only for the forecast, even when there is no demand. Certain quantity can be push then onwards it can be the pull-based strategy.
We need to strike a balance between the strategies.
- September 22, 2016 at 4:06pm
Thank you all for your nice comments.

@ peter: Agreed changing the part from pull to push is nontrivial. Nonetheless, In RR you can turn on the push planning for a part. The logic for a part to be pushed is based on the demand profile. It means, if the inventory reaches the maximum level and there is a demand for the part at down stream site then it is pushed. This logic provides more flexibility compare to when the part is pushed according to it's target stock level no matter if there is demand for it.

@Krishna: Totally agreed. The difference between the historical actual demand and forecast (forecast error) comes handy if there is more need for push, specially if the part has a high seasonality demand.
- July 14, 2020 at 10:44pm
Thank you! Your explanation and use or example made this easier for me to understand. It's a topic in an exam I am sitting soon and was looking for a clear simple explanation.

Leave a Reply