As many of us would have observed, forecasting often cops the blame for many issues within the supply chain. This has only been compounded by the proliferation of terms such as “demand driven supply chains” increasing awareness of the importance of demand planning. However, the inference that forecast accuracy is the sole cause of issues is dangerous and stems from two main misconceptions.
Firstly, forecast accuracy should not be considered as an end in itself. Instead, it is one contributor to outcomes that needs to be managed to deliver expected results.
Secondly, despite of all best attempts, forecasts by their very nature will never be completely correct. It’s much more important for your organisation to recognise this fact and manage an understood level of risk than to expend too much effort improving forecasting when it may not translate to better outcomes.
What this means for your organisation is that in an environment where increased volatility, shortening product lifecycles and complex customer fulfilment models are making forecasting increasingly difficult, the effort required to improve forecasts may not provide the expected supply chain benefits. This is particularly true if you have broken links in your supply. For example, increased volatility will be reflected in increased forecast error – and this should directly translate to the quantity of safety stock required. However, if you are using a blanket safety stock policy such as weeks cover, you won’t be able to benefit from or buffer changes in forecasting performance or risk profiles. To put this in an everyday context, if you were to move from a house where you park your car on the street into a new house with a garage, and yet you retained your existing car insurance policy, you would not be benefiting from the lower risk of your new parking arrangements.
Either way, it’s more important that your organisation understands this volatility, can anticipate the expected error of statistical forecasts, and is able to manage this risk. In fact, there has even been discussion that given the operating environment that many organisations face, forecasting has little relevance to the short-term, executional horizon and should focus on buffering this variability while preserving demand plans for longer-term capacity and strategic planning.
While such an approach may not be appropriate for every organisation, the question that should to be asked in relation to this is how right does your forecast need to be? That is, what level of aggregate risk is manageable given your customer offer and working capital constraints? The balance between improving forecasts and managing risk via responsiveness (including the use of inventory) should flow from your organisational and supply chain strategy. In order to assist this process, many organisations use an inventory optimisation tool to model customer service targets against varying levels of forecast error and working capital. If an acceptable mix cannot be found, you may need to revisit your Customer Offer. Taking this approach to setting forecast accuracy targets will provide a more beneficial outcome for the organisation than, for example, taking last years’ number and increasing it by x %.
However, this is not necessarily an easy process – managing these trade-offs requires a mature, organisation wide dialogue. Without this understanding, it’s impossible to understand the ROI of forecast improvement initiatives, especially as efforts to improve forecast accuracy may suffer from diminishing returns.
Forecasts are just one driver of performance. Without a clear understanding of the expected level of error within your demand plan, plans cannot be built that manage this risk. However, if this risk is understood and the management of this risk reflects current strategy, your organisation can drive consistent, beneficial outcomes.