Goodbye forecasting

If we look around in nature, there are several super-efficient supply chains which run day in and day out without the need for forecasting. Back home, our moms run the kitchen supply chains with hardly any stock-outs or obsolescence. However, when it comes to packaged consumer goods firms, even the highly efficient ones seem to depend a lot on forecasts.

In the FMCG industry, it’s a fact that consumer tolerance time is much shorter than the response time to plan, source, make and deliver. Is that the reason why we have to rely on forecasting? How long should the planning horizon be to run an efficient supply chain?

As we improve the agility of supply chains, the dependence on forecasting indeed comes down. More agile supply chains can do with shorter planning horizons. The most agile systems for example - the ones found in nature, can respond in real time without the need for future visibility. Is the length of planning horizon, therefore, a reflection of how slack our supply chain is? How much more agile do we need to become?

We have come a long way in improving the accuracy of forecasts by segmenting the product portfolio and applying a plethora of sophisticated statistical models. My compliments to the brains behind these models! However, the goal post itself seems to be moving and we find it difficult to go beyond 80% forecast accuracy at the granular level of product-geography combination.

In my opinion, since forecasts are anyway going to be unreliable, the preferred direction of solution for consumer goods companies should be two pronged… to improve agility and to reduce forecast dependence on an ongoing basis. Of course, the final objective remains unchanged... to become so agile that we don’t depend on forecast at all.

Converting demand signals directly into supply actions requires a carefully orchestrated process design with inventory buffers placed at strategic points. These inventory buffers should be dynamic in nature. Their size depends on replenishment lead time from the previous node and varies with velocity and variability of consumption. Concepts of Theory of Constraints (TOC) and Demand Driven Material Requirements Planning (DDMRP) are of great help in designing such a process.

Apart from a good supply chain design, organisations need to improve their flexibility in sourcing, manufacturing and delivery using SMED principles as well as the TOC concepts of Focusing and Protective Capacity. They also need to improve their response time through flawless execution and suitable exception monitoring systems. These are pre-requisites for delinking supply actions from forecast. I have seen organisations struggle with TOC and DDMRP implementation if they have not focused on these pre-requisites in parallel.

With a proper supply chain design and essentials of flexibility and fast response in place, it is indeed possible to run the consumer goods supply chains without forecasting and synchronise supply with demand in a dynamic manner on an ongoing basis. Successful implementations of TOC and DDMRP have seen customer service levels jump from nearly 80% to as high as 99%.

Moving from forecasting to demand-driven supply chain is especially relevant now because consumer tolerance time will keep coming down in future. I will elaborate more on the transformation process in my future articles.