Corporate forecasting based on sales trends is not good enough. Retailers must create synergy between corporate forecasting and store-level merchandising strategies.
Wal-Mart has proven to us that the proper relationship between forecasting and merchandising can give retailers the supply chain edge they need to beat the competition. Here, Henry Bruce, VP of strategy & marketing at IMI Americas (Mt. Laurel, NJ); Tom Nowak, senior VP of IS (information services) and CIO at Price Chopper; and Dennis Krautsack, senior director of IT at Dollar General discuss the attributes of a successful corporate forecasting/store merchandising relationship.
1. What are the attributes of a flexible retail supply chain model?
Bruce: The model is event driven. They are basically taking the tenets of demand-based replenishment and putting them in place at the store level. You have department managers that have complete access to POS information, as well as trend information based on past performance of promotions. Dealing with the normal turn business of goods that have a steady stream of demand is not the issue. The issue is dealing with direct store delivery, seasonal sales, and promotions. That's where the out-of-stock problem is exasperated.
Krautsack: Influxes of demand for specific products can be handled any number of ways. Sales must be analyzed often to determine trends. But, more importantly, our managers have the ability to order merchandise. Even though we have automatic replenishment systems in place, the store manager has the autonomy to order more, or order down, if that's necessary. They can adjust the replenishment parameters. Of course, there are exception-based systemic controls in place, so when stores place burdens on our inventory by dramatically increasing orders, we get a report on it.
Nowak: In the grocery industry, we have large orders taking place every single day of the week. Unanticipated influxes of demand are not as much a problem for grocers as they can be for general merchandisers. It's not practical for us to track sales trends hourly, but we could if we needed to. There's an order cycle every evening in most stores, so we can react to low inventory within 24 hours. Before we had a system, it was the experience of the order writer. With systems in place, you get a better forecast and spikes can be anticipated.
2. What integration issues arise in the merchandising/forecasting relationship?
Bruce: It's all about integrating POS information with in-store operations, and tying that back to headquarters for promotion planning. I need to know what's going off the shelf and why. Is it an end-cap display? Is it a promotion? If I understand why a product moves, the way in which I forecast for my store and pass that information to the ordering clerk becomes a whole different scenario. Instead of forecasting from the aggregate, I'm forecasting at a very granular level. I'm forecasting sales of a particular item selling off of a particular shelf in a particular store. It's a bottom-up approach.
Krautsack: We have an annual line review. During the line review, we look at an item's sales figures gathered from the POS in relation to its product mix and visual presentation. Then, we look at market indicators. We bring in data from several outside consultants and conduct a sales analysis. This helps us establish whether or not we're attaining our market share. Out of that process, items are either dropped or added, and new planograms are developed.
Nowak: Perpetual inventory is an integral part of the grocery system, so you have to integrate with your POS information as well as any of your billing systems. In grocery, we get some of our product from our DC (distribution center), and some that's delivered directly to the store. There are two billing systems right there we have to provide links to. We also have to link to our warehousing and fulfillment systems back in the DC once the orders are generated.
3. Shed light on the balance between corporate and store-level supply chain decision-making.
Bruce: If you think about what's going on in terms of analysis at corporate, they're looking at promotions weeks in advance. As the event gets closer, there should be communication with the store. For example, if I'm planning a promotion for a store, the store is the one that knows that lifts in sales happen because of certain end-cap or aisle placement. The ability to share this knowledge is where the difference is made. The corporate office needs to know that sales performance has as much to do with store merchandise execution as it does with corporate forecasting. Merchandising variables are what makes the difference between a 3% to 4% out of stock versus a 5% to 8% out of stock situation.
Krautsack: For our core merchandise, we have fairly rigid planograms in place. But, the stores have leeway with some end-caps and wings. Seasonal and one-time impulse buy items aren't planogrammed by item. We simply allocate shelf space and give the managers latitude to display merchandise creatively and in a manner that will maximize sales in their local community.
Nowak: Normal turn merchandise should be as automatic a process as possible. Previously, we relied on the experience of our order takers, but that became problematic as we grew, as turnover increased, and as people got transferred around. Turn merchandise needs a systems-based forecasting solution. On sale merchandise, initial forecasts are done back at headquarters based on past sale history. There's typically a large order of sale merchandise sent to each store prior to the sale, called a 'beginning of sale distribution.' But then the store's system takes over and does the remainder of the forecasting, based on the first few days of the sale. This provides a more accurate picture of anticipated merchandise movement over the whole term of the sale.