The human touch in forecasting and demand planning

This is an excerpt of the original article. It was written for the September-October 2021 edition of Supply Chain Management Review. The full article is available to current subscribers.

September-October 2021

This time every year, we publish Gartner’s Top 25 supply chains, the annual list of the supply chains that have made it to the top, a list that now also includes 5 Masters, or companies that have consistently outperformed year after year. You can read the article in this issue, along with the web exclusive material we publish on scmr.com, to find out what it takes to become a supply chain leader.

Browse this issue archive.

Already a subscriber? Access full edition now.

Need Help?

Contact customer service

847-559-7581   More options

Not a subscriber? Start your magazine subscription.

Much attention is being paid to the various methods by which the computer-aided automation of—primarily—autoregressive forecasting can be undertaken. And with good reason. Artificial Intelligence, Machine Learning and a host of other technologies represent remarkable developments in the forecasting world. They hold the promise of better leveraging the efforts and insights of demand planners at learning what their order history can tell them about the future.

For SKUs with a particular demand profile, there is little question that forecast automation may yield a significant improvement in both performance and cycle time, provided there are no substantive changes in demand in the future. Nearly all businesses, however, have items whose demand profiles do not lend themselves well to autoregressive forecasting, and there is no business that can’t benefit from the judicious application of human input under the appropriate circumstances. Leading forecast researchers Paul Goodwin and Robert Fildes have demonstrated this over long periods of observation.

Whatever the profile of a company’s demand, and whatever its degree of automation, there are nevertheless multiple touchpoints where human intervention can consciously or unconsciously transmit biases into the demand planning process. Indeed, my company, NorthFind Management, worked with a number of multi-billion-dollar global manufacturers to understand the prevalence of biases and heuristics in the general population within the organization, and specifically in demand planners.

How biases influence planning

From the inception of demand planning, unconscious biases and heuristics have influenced the sources of data that planners consider or exclude as part of the demand plan. For example, if the prevailing wisdom in an organization holds that their business is unique—a sentiment held by most companies—there may be a reluctance to invest in mining syndicated channel data for additional insights into demand, despite the probability that, when properly used this source of data yields benefits. In this case, two prevailing biases—Availability Heuristic and Groupthink—unconsciously bias the demand planner and, as a result, directly influence the entire planning process.

SC
MR