Today‘s supply chains are undergoing significant pressures to become more “demand-driven”. Retailers, distributors and manufacturers are forced to choose the approach they hope will make them the most profit. Is it producing and making goods available to forecasts of expected consumer demand, or by reacting to what consumers have already bought? In this post and in the next, I want to show why companies are investing in a new generation of cloud technologies that enable the transition from “push” environments to “pull” environments. But first, what is the difference between a push and a pull system?
What is a push system?
Most companies use the forecast approach today, in what is called a “Push system”. Companies forecast to feel confident that the goods they order will both find willing buyers and not run out unexpectedly soon. In the Push world, decision points occur at every reorder. How much should be purchased? In other words, how often is it necessary to consider buying each item?
The sheer volume of SKUs and associated decision points mean that push systems use the “peanut butter approach”, where all products are treated roughly the same despite their different demand profiles. Thus we see the following:
- Forecasting done at the aggregate level. This is a weekly forecast at best, and is from the DC to the store.
- Product is then pushed to the store weekly based on this forecast with no granularity based on how much the individual SKU is actually selling for that particular store.
The main problem with push systems is that they are based on forecasts that are almost always wrong. Despite billions spent annually in the US for the best computers and most sophisticated software, actual demand varies from forecasts. Forecasting does not make the end consumer react more rationally or predictably. When it comes right down to it, no matter how sophisticated its algorithm, a forecast is only a guess. Wrong guesses mean excess investment and lower profits, due to missed sales. They also lead to other problems like high carrying costs, discounting, disposals, missed sales, weak customer loyalty, shortages, high debt loads, inventory disposals, emergency shipments, rescheduled production and attenuated profits.
What is a Pull System?
Modern cloud-based technologies are enabling a true pull-based approach to retail replenishment that uses actual daily consumer-level demand to generate a true forecast.
- Pull systems use demand data to drive both replenishment and production. Only immediate customer requirements are drawn from the protective inventories upstream.
- This approach is driven by actual consumption at the store (store/SKU/daily demand with POS) as well as with forecasts. This allows for a much more granular approach than push systems.
- For example, a pull network supports multiple replenishment policies based on the individual demand profile of the product.
- E.g. for a “slow mover” you can manage by a simple reorder point (sell one, replenish one).
- For a turn item however, you can use a more sophisticated min/max policy with DOS.
- The result is an automated inventory policy driven by actual pull requirements at the granular level. By acting on actual demand, statistical variations are damped rather than magnified, steadying on-hand inventory levels at every stocking location.
- Since goods only flow downstream to cover immediate need, the preponderance of the inventory remains further up the supply chain, closer to the source. In contrast, many push systems put the majority of the inventory at the retail store.
Next, I want to talk about the advantages of pull systems, but if you can’t wait then I suggest you download and read the new Push vs. Pull Tech Brief.
Latest posts by Aaron Pittman (see all)
- The 4PL Revolution Roundup: Top 8 Opportunities for Today’s Logistics Providers - December 15, 2015
- Demand Sensing Round-Up (Blog Posts, White Papers, and Webinars!) - December 15, 2015
- Your Favorite Posts of 2015 - December 4, 2015