Retail Acumen is the new name for ARC-Retail.

Please visit our fantastic new website at

www.retailacumen.com

Lies, damn lies and statistics - that's the basis behind demand forecasting. Various methods and volumes of data, algorithms and calculations. In the end though only one thing is certain. The forecast is going to be wrong!

For retailers, understanding the measure of the inaccuracy and being able to provision for the risk of out of stock or over stock as a result of that inaccuracy, is key to unlocking the supply chain opportunity.

Once the products are signed off from the range and space planning processes the demand planner really comes into their own. With good information about the market, performance, demand influencing factors and comparable products, a skilled forecaster can estimate the rate of sale and seasonality of a new product introduction with a fairly high degree of accuracy - adequate for starting negotiations on price related to volumes at least, not to mention for populating distribution capacity plans.

The complexity really only arises as the product is received into the retailer's supply chain and requires to be further distributed to stores where it can be sold. Questions around which stores will perform best for which products are asked, and at this level of disaggregation the data tends not to be as statistically valid for the forecast algorithms as at the higher levels of aggregation. Achieving an accurate sku-store level forecast, then cut further down into weekly, even daily, time buckets, would be the panacea for a demand planner. It would give validity to the level of display stock set into the system, it would enable push back over decisions taken surrounding the store assortments and it would give a much better ability to run a completely pull based replenishment-to-order-up-to-level driven supply chain.

Arguably demand forecasting is one of the most data hungry processes in retailing, where data accuracy (or not) has massive repercussions and where simple errors in processes such as a store physical count, an inaccurate order lead time etc can actually result in a store residual stock problem which can erode the net delivered margin of a product significantly in just one inaccurate user's key-stroke.

However, demand forecasting need not be the poisioned chalice of business processes. With clear processes and quality data, retailers can usually significantly benefit from improved forecasting capabilities.