Computer generated ordering: Taking the guesswork out of store-level replenishment
This article was previously published in Retailspeak magazine.
When Mr. Brown enters his local supermarket with a shopping list of nine items, he has less than a 50-percent chance of finding everything he came in for. And when Ms. Parker, vice president of inventory management in a mid-size retail chain, analyzes the number of inventory turns in the chain for the last 10 years, she probably won't discover any consistent improvement over that time.
Surveys show that on average eight percent of all items are out-of-stock at a typical supermarket with hardly any improvement over the last decade. Surprisingly, inventory turns have been relatively stable for the last 10 to 15 years.
Retailers are well aware that one of the key components of a successful operation is store-level replenishment. Forty-seven percent of out-of-stock scenarios in stores are due to inaccurate store-level ordering decisions. These stock-outs result in a four-percent loss of sales and can have a negative impact on customer satisfaction and loyalty.
At the same time, failure to reduce inventory levels prevents retailers from freeing capital for expansion. It also creates additional costs due to higher rates of spoilage of perishable items and higher operational costs due to moving items from the packed back room to the shelves.
Tools for efficient store-level replenishment help retailers to greatly improve their efficiency. These systems help store associates, who typically generate orders based on their intuition and experience, to use scientifically generated order recommendations that take all relevant data into account.
When working properly, such systems result in improved service levels and customer satisfaction, reduced inventory levels, reduced spoilage of perishable products, and fresher products.
One of the key elements that defines and differentiates a successful store-level replenishment system is a computer generated ordering (CGO) solution.
The science
A computer generated ordering (CGO) system needs to provide an accurate forecast and an optimal order based on that forecast.
Forecasting consumer demand at the store is a complex task. The biggest difficulty is producing a forecasting algorithm that can identify real trends, seasonal patterns, and lifts generated by recent promotions, rather than wrongly identifying normal demand fluctuations as trends and patterns.
For example, if Mr. Brown shops somewhere other than his regular store, he may double the daily demand of his favorite flavor of yogurt in that particular store. In this case, the successful CGO system can determine whether this is a temporary demand shift because Mr. Brown happens to shop there on that day, or whether the store is experiencing a general increase in popularity for that particular flavor of yogurt, which will be sustained over time.
Calculating the optimal order is an overlooked problem in retail store-level replenishment. Many retailers and software vendors believe that manually calculating a desired service level will do the trick. However, these service levels cannot be sustained with a manual or even semi-manual process. This will invariably result in an army of people devoting themselves to maintaining a suboptimal replenishment system.
A good CGO system should automatically determine the optimal service level for each item, in every store, and should change on a periodical basis. This way, retailers can optimize their stock levels in keeping with customer demand.
The process
Keep it simple: The computer generated ordering (CGO) system interacts with store associates who can override the recommended orders, make inventory adjustments, and perform scheduled inventory counts. A good CGO solution must include easy-to-use handheld technology that can support the workflows and data needs of the store associates.
Focus on exceptions: You can maintain an accurate inventory by assigning 10 associates to walk the floor and count inventory all day. But the more efficient way of doing it is for associates to perform smart counts. This involves counting only subsets of items that will be pointed out by the CGO solution, such as those with high probability or with a high shrink rate. As a result, retailers can make immediate labor and inventory cost savings.
You could require your associates to review the entire list of recommended orders every day. But with a good CGO solution in place, they only need to review the exceptions, such as items where the recommended order deviates from the normal. By doing so, associates can save time and money.
Deliver corporate control: Retailers need to improve key performance indicators (KPIs) such as service levels, number of turns, and percentages of out-of-stock scenarios. Managers must be able to monitor store performance, not only by using KPI reports, but also by getting alerts from the CGO system. With the solution in place, managers can receive alerts such as Store 79 service level dropped below our target, Store 67 is making too many changes to orders causing increased inventory, and Store 137 is not performing smart counts and the inventory accuracy level is deteriorating.
Deploying computer generated ordering (CGO) solutions with high scientific quality combined with process workflow support can take the guesswork out of store-level replenishment and maximize store profitability.
