How to improve demand forecasting is an area of intense interest with our readers, with blogs on forecasting consistently ranking as the most popular. So we’ve picked the five best read blogs on how innovation, technology and business strategy impact demand forecasting transformation. Below are links (in the titles) and summaries to each of these five blogs. We hope these shared insights will help you on your journey to improved supply chain performance.
In 2016, it seemed as if everyone was looking to improve their forecasting performance. In this blog, Ben YoKell, who oversees Chainalytics’ Demand Planning Intelligence Consortium, asks the question, “How much can you realistically expect to improve your forecast accuracy each year? And how do you set your targets for the coming year?”
According to Ben, many companies take last year's forecast accuracy metric and simply add a few percentage points to establish the coming year’s goals. YoKell outlines a new alternative to relying on a historically-anchored approach. Rather than setting forecast accuracy targets by negotiation from historical performance, he shows how real empirical data can define what is realistic or even possible for the nature of the demand being forecasted.
This blog describes a six step model to more powerful and accurate demand forecasting. It includes stages such as naive forecasting, statistical forecasting and demand modeling, and highlights the difference between each. Gartner would call this a “maturity model”. The blog includes a popular Evolution of Forecasting infographic which shows each stage in the evolutionary process.
This was a joint effort by myself and Stefan de Kok, a widely read blogger with a strong academic background in supply chain planning. We challenged readers with the fundamental question, “Maybe you are doing your demand forecasting completely wrong. To be more precise, there are two equally important outputs of demand forecasting and you may be focusing nearly all your energy on only one, and maybe even the wrong one.”
If so, the impact is that you may not be getting the forecast accuracy you want. Or even more important, that you may not be getting the service levels and inventory efficiencies that you need. And if that’s true, you are not alone. The number of companies is growing that are saying that their forecast accuracy, service levels and inventory efficiency metrics have hit a ceiling that they just can’t get past. We look into the root cause of the problem and offer readers a way out of the dilemma.
In this blog we explain the difference between forecasting demand and modeling demand. Forecasting typically starts with an aggregated time series of data—usually presented as a bar chart displaying demand one period after another. Based on what's happened in aggregate over the last months or years, it makes a projection of what will occur in future months. But when you aggregate you lose signal; signal that can never be retrieved again at the aggregate level. You trade away accuracy for apparent ease.
Demand modeling works the opposite. It breaks the demand components into a series of internal and external factors—the demand stream—and looks at how each impacts demand to predict future demand at a granular level for individual SKU-Locations.
We have written several blogs about how to incorporate machine learning in the demand forecasting process, including topics such as promotion forecasting, new product introductions, extreme seasonality and even “social sensing”. This is our kick-off blog on the topic and the genesis of a longer article in CSCMP Quarterly magazine. It describes how machine learning refines the demand model to reliably overcome the many causes of demand variation that produce significant forecast errors.