Slow and Intermittent Product Demand Forecasting Myths
Test the Slow and Intermittent Product Demand Forecast Myth
Reviewing several retail software provider web sites, a visitor will see that many software companies highlight their ability to forecast slow or intermittent demand products. Following that line of reason, the measure of success is found by testing the forecast accuracy for a given period of time and repeating the test regularly. Taking the results and calculating the mean and standard deviation will tell us if the forecast is accurate. The problem is that forecast accuracy is not the ‘smoking gun’ solution to slow or intermittent demand products.
To manage slow and intermittent product demand, we recommend you avoid:
- Reviewing forecast accuracy within a single company traditional demand period
- Measuring in-stock
- Using incorrect forecast methods like Time Series, Exponential Smoothing, and This Year/Last Year. These fail to deliver the correct stock model. Some of these smooth or ignore zeros, which is a bad idea. Croston’s method may be used, but, without thoughtful implementation, then it, too, will deliver poor results.
[QuickLinks]When managing slow and intermittent product demand, you should:
- Review forecast accuracy over a window of time that includes multiple periods.
- Measure service attained – NOT in-stock. The focus needs to be on the inventory stocking requirements necessary to meet service level.
- Use inventory management software that correctly identifies slow and intermittent demand. This software should not smooth or ignore a zero in sales history when calculating a forecast. The software should deliver an inventory management solution that is based on service level and cost-effective inventory management models.
Service Attained = [Sales / (Sales + Lost Sales)]
How to Achieve Service: Optimize Facings and Ignore Demand Forecast
The pick size and store facing (presentation model) often have a significant impact in the end results. If a product has a forecast of 4 units a quarter and a pick size of 4, then the forecast accuracy only really matters across 3 months (1 quarter). If the stock model or store facing is one unit, when the on hand is one unit, then the system will buy more. In this store facing example, the forecast accuracy again only matters across 2-3 months. The critical metric to look at here is service attained and not in-stock. To put it differently, inventory to meet customer demand and sales results is what’s important.
Overstock, Shelf-Life and Smoothing the Zeros
Without a good strategy for slow and intermittent product demand, there will always be overstock. The resulting overstock will force you to commit more inventory dollars to these products that could be used elsewhere. Shelf-life issues will also be created by smoothing through the zeroes in the sales history. The pains of overstock and inventory lost due to shelf life expiration can occur at the same time without a strategy for slow and intermittent product demand.
Focus on a Combination of Service and Demand Forecasting
The key to managing slow and intermittent product demand is to focus on a combination of service and demand forecast accuracy across a wider selection of periods to avoid the common pitfalls listed. Identify the slow and intermittent demand products in your assortment and try some of these ideas for the next 45 days to learn how this helps to ‘Tighten the Links in Your Chain™.’
Are your ready to ‘Tighten the Links in Your Supply Chain?™’
We are here and ready to help. Contact us for a free consultation about your forecast accuracy and inventory management opportunities. You can also request a demo and see how things can really start to improve in your business in 90 days.
Copyright © Data Profits, Inc. 2013 All Rights Reserved.
- Lead Time Forecasting - February 7, 2019
- Data Profits in Gartner’s 2017 Retail Replenishment and Forecasting Software Guide for the Sixth Year - February 13, 2018
- Data Profits Releases 4 Easy Replenishment Ideas that Adapt to the Digital Age - July 20, 2017