My career spanning almost 40 years was to manage catalog inventory for a major Retail company selling apparel, shoes, home furnishings, toys, and sporting goods. The major principles learned in catalog inventory management also apply to stores and online sales, though store and online businesses have greater flexibility than a catalog in terms of reacting to sales increases and decreases from pre-season item estimates as they can instantly lower prices on poor selling items to boost sales and also replace out of stock items with other items so that the customers don't see the out of stock items. You can't do that with a printed catalog page within catalogs at customer homes.
What applies to all forms of retailing is that the total average sales of any item when measured to the total customer base that sees that item is statistically insignificant (roughly about .0001 of the customer base so that the “typical” average item sales of a customer base of 10 million customers is about 100 units sold). “Average” is calculated from the total number of items sold and there will be many items that sell significantly more than average plus many items that sell significantly less than average. If all items sold at the average number, there would be no problem forecasting and stocking the item, but that is not the case.
Keep in mind that many items come in multiple sizes and/or colors. It is at the size color level that stock levels need to be accurate (called the sock keeping unit or “sku” level for short) and sku level sales are much, much less than their item level sales. Furthermore, the sku level inventory needs to be broken out by geographical locations of multiple stores and distribution centers/warehouses. There can be significant sales differences and sales rate differences among items and skus in different geographical locations.
Due to the statistical insignificance of item and sku sales, the forecast accuracy of a pre-season item forecast is dismal. Only about 15% are accurate to plus or minus 10%. Therefore, 85% of all pre-season item forecasts are inaccurate! Given the fact that once sales start trickling in for an item, since those sales are far less than it's seasonal estimate, the statistical accuracy of those forecasts is often horrible.
There is no way that a Retailer can or would stock 100% of the pre-season estimate for all items before the selling season begins for multiple reasons. If a retailer did that, about half the items would be over bought resulting in very high markdowns to get rid of the inventory , thus hurting profits. Keep in mind that without profits, a retailer would go out of business. Also, the inventory space requirements needed to do that would be astronomically high and expensive, hurting profits plus resulting in higher prices in order to afford to do that. Therefore, “turnover” (the multiple usage of a fixed space to store an item repeatedly through multiple seasonal purchases) is extremely important for all retailers if a profit is to be achieved.
If the item being sold is imported, the retailer may have little choice but to buy 100% or close to 100% of it because many imported items have only one manufacturer production run (or at most two production runs within the same season) whereas the manufacturer of a domestic item may have several seasonal production runs enabling retailers to make multiple purchases at different times for different ship dates, thus facilitating good turnover and also increased forecasting accuracy over time for domestic items. Thus both overstock and out of stock situations are more common for imported items.
One other important factor resulting in imbalances of inventory are item “pre-packs”. A pre-pack is a manufacturer's shipping package with pre-determined units by sku. Compared to catalog and online orders to manufacturers, individual store orders are usually much smaller because an individual store's item sales are much smaller as their geographical “reach” is their own “local” area, not regional or national as would be the case with catalog and online service areas. For example with overstock on one sku and understocked on another sku for the same item would not be buying enough for the understocked sku and buying too much for the overstocked sku when ordering pre-packs. Some store chains mitigate this problem by buying to regional distribution centers and servicing their stores from them. With national catalogs and online businesses, their orders are large enough so that manufacturers allow them to buy their individual sku needs, not pre-packs.
The best inventory management defense for a retailer is to have the people responsible for forecasting and purchasing item inventory be good “merchants”. In the beginning of the season when item sales are a tiny portion of their final seasonal sales totals, a “random” insignificant difference in sales to date can, when measured against historical rates of sales for that same time period the year before (maybe less than 5 or 10% done or especially less than 1% done) give astronomical mechanically generated forecasts that are highly unlikely. A good “merchant” considers factors beyond statistically generated historical forecasts based on few sales (that are highly likely to be very inaccurate), and tempers any item estimate revisions accordingly so that drastic mistakes are not made when it is impossible to be statistically accurate.
Bottom Line – We call it Inventory Management but in reality, it is “Risk” that is being actively managed because item and sku forecast accuracy that is needed (within the time frames of each individual manufacturer production runs) for good inventory management is pretty much an unattainable goal. That is why Retailers, no matter how much they work to keep in stock on all items and skus, will never fully achieve that goal.
