By all means use forecasting time fences. And by all means violate your time fences.

A time fence is a defined period in which the forecast should not be changed. An item's forecasting time fence often mirrors the item's planning or sourcing lead time. If it takes two months from when an item is ordered to when it's received into inventory, there's logic in not changing the forecast within that two-month period, since a change to the forecast will have no impact on the item's availability. In that case, you would put a two-month time fence in place. 

Forecasting Time FenceSome companies impose "hard" time fences (in which the forecasts absolutely cannot be changed), while others enact "soft" time fences (in which the forecasts can be changed – as long as the changes are significant and well-justified).

A time fence provides planners with assurance that the forecast will not change within the item’s lead time. Is this necessarily a good thing, though?

One school of thought holds that since supply plans cannot be changed within the time fence, the forecasts should not be changed. Another holds that since supply and demand must be balanced over time, it’s foolish not to reflect significant changes in expected demand in the forecast – even within the time fence. A forecast change within the time fence won't impact an item's near term availability, but it will be reflected in the item's longer term projected inventory.

While we understand that planners will not be able to revise their sourcing or production plans within their lead times, there is still value to providing them with forecasts that reflect significant changes in expected future demand… since that information will ripple through and impact their plans on the other side of the time fence.

Looking at it another way, a significant forecast increase or decrease within a time fence won't affect your most recent order, but in all likelihood it will have a heavy impact on the timing and the quantity of your next order.

We suggest that you employ time fences to help ensure discipline and accountability. Forecast accuracy measurements should be based on the forecasts just outside of the time fence. At the same time we also recommend that you update forecasts within the time fence when you are highly confident of significant changes in demand. The result will be a more dynamic and realistic set of forecasts and plans.

(Note: We're well aware that this view on time fences is not unanimously held. We invite your commentary, your dissent, as well as anecdotes about your experiences with time fences).

This article is excerpted from the e-book "42 Ways to Improve Your Forecasts" which will be published in May.

_____________________________________________________________________________________________________________

Bill Whiteside was introduced to Demand Solutions in the late 1980s when he purchased the software for a dairy products company (his employer at the time) in Lancaster, PA. On Jan. 1, 1990, he made the ultimate product endorsement by starting a business to market and support Demand Solutions in the Northeast U.S. In his sales, marketing, and support roles, Bill has worked with more than 400 companies across a diverse group of industries.  He writes and speaks frequently on forecasting and supply chain planning topics. Download and read Bill's "12 Supply Chain Forecasting Lessons" white paper today.