RETAIL INNOVATOR: That's an order
It's difficult to hand over complete control of an operation to a computer. One somehow envisions Skynet of "Terminator 3," complete with nuclear missile attacks aimed at Earth's major cities. While supermarket executives don't fear total annihilation from completely automating their ordering processes, they tend not to be too comfortable about what might happen once they hand over the reins to machines.
Such was the case at United Supermarkets, Ltd., a 47-store chain based in Lubbock, Texas, which is in the midst of implementing a computer-automated ordering (CAO) system. "With any initiative like this, there are a lot of people in the operation who get a bit nervous," says United v.p. and c.i.o. Scott Gilmour, who spoke this past March at the FMI Marketechnics session called "Building a Knowledge-based Computer-Automated Ordering System." "Chief financial officers, for example, get nervous because you're changing to perpetual inventory from -- in our case -- manual accounting three times a year."
Automated ordering is just one of the many projects occurring under Gilmour's watch. United's technology infrastructure is divided into three areas: corporate systems, store systems, and supply chain/logistics systems, and Gilmour's staff is currently involved in various implementations in each of these, among them establishing an asset management system, integrating the fuel point of sale with the store point of sale, and building a Web-based labor-scheduling system.
Some of the pilot objectives Gilmour had for the CAO system were to reduce the number of out-of-stocks, automatically detect when an out-of-stock occurs, optimize the shelf stock by decreasing storage costs and inventory, reduce handling costs, and reduce ordering and inventory management labor.
Gilmour chose to use SofTechnics and SAF AG for the project. "We were the first grocer to install this technology from these companies," he says. "We wanted to help them to demonstrate and validate the solution, to evaluate the system's performance and measure the results, improve the capabilities, and identify product enhancements."
Vendor goals
SofTechnics and SAF, on the other hand, were looking to gain valuable project experience and to help ensure United's success during the technology companies' first venture into food retail. "We wanted this to be a win-win for both of us," Gilmour says. "We were also looking to achieve feasibility study key performance indicator results regarding out-of-stock reduction, and we wanted to reduce our inventory to optimal levels."
This wasn't going to be an easy task. The difference between computer-assisted ordering -- which is what United was doing -- and computer-automated ordering is that in the latter there's no human intervention. In computer-assisted ordering a suggested order is created by computer, and an employee is prompted to accept the order or make changes. In computer-automated ordering the computer creates and generates the order; it doesn't prompt someone for approval.
This means that if the store's inventory isn't accounted for, then the ordering system won't do its job. "There are a lot of disciplines needed for this," Gilmour points out. "Everything from 100 percent scanning to transfer from one department to another, for transfers to other stores, to reclamation, to throwaway, all those things have to be scanned. Obviously the detailed invoicing at the back end is the starting point, but there are a lot of disciplines that need to be put in place."
United needed to make some upgrades before it began its feasibility study in the summer of 2002. Gilmour's team had to upgrade its system from an earlier version of SofTechnics to a newer one and had to convert from DBfile to a more industrial-strength relational database, which was found in Informix.
Gilmour already had a jump on the second step, which was to create detailed invoicing. "We were already in production with detailed invoicing from our warehouse," he says. "The fact that we started with dry grocery items and frozen items -- for which we were already receiving detailed invoices -- was a major hurdle which we had already cleared."
Personnel issues
The next step was crucial and could make or break the project: getting the appropriate personnel into the two pilot stores and making sure they were trained to implement a process to scan all transfers and activate scan data collection.
The forecasting engine for the CAO system required two years of previous data for the feasibility study and pilot initiation. While United had detailed sales data for the previous years, it didn't have any promotional history. To address this problem, SAF developed a promotion detection script based on assumptions about sales spikes.
The last step was to count all of the product the evening before the system went live. "We had to physically count all of the items to initialize both the physical inventory quantities and to set up items for automated replenishment," Gilmour says. "There was some obvious labor involved with this. We needed an accurate count of what we had in the back rooms and on the shelves and on the displays for each item, and then we turned on the system after the count was completed."
Gilmour started with one store, then began the other pilot in a second store a week later, implementing the same process in each.
Looking back, Gilmour would have changed one item if he'd had the chance, and he recommends that any retailers undergoing such an implementation take note: "We decided upon major product lines, like paper products, canned fruits and canned vegetables, cereals, pet food, and frozen food," he says. "The one thing we didn't take into consideration was the opportunity to reset these areas. Later, after the initial pilot, we turned off a few of these products and did facing aisles so we could bring in our category management staff and employ our ACNielsen Spaceman shelf management system to do some resets."
Still, the results were encouraging. In one store, inventory was reduced 17 percent, and the other saw a 22 percent decrease. The value of the inventory was decreased 14 percent and 12 percent. Out-of-stocks dropped a whopping 58 percent in one store, and 62 percent in the other.
There was an increase in sales for each store undergoing the pilot, but Gilmour takes this with a grain of salt. "We can't say that this is just from the CAO system," he says. "There may have been a number of other things involved in our ability to increase sales."
Interestingly, the actual results were very close to the simulation that was performed prior to the start of the pilot.
Reduced labor
One of the goals of implementing the CAO system was to reduce the labor required for taking inventory. "We don't want to have somebody go around and physically count these items," Gilmour says. "We want the system to do it through the detailed invoices coming in the back door and through effective scanning of reclamations and transfers. This should be no problem if the system is tracking our inventory accurately."
During the pilot stages Gilmour found that only 6 percent of the items required manual intervention. "This was due to alerts that we set up in the system, such as too many zero sales on an item, a new listing, zero inventory, or a drop in sales," he explains. "An employee who's working in the store would count the item that was affected to keep the system accurate."
Any change to fully automated tasks can't be completed successfully without addressing issues of change management, and this was a topic Gilmour made sure he addressed upfront. "We wanted to empower our associates with an understanding of the benefits of the CAO system," he says. "The emphasis on the importance of inventory discipline was critical."
He also had to address the issue of resets. "When it came to resets, [associates] had to get used to the idea of facing rather than filling -- at least initially. Our stores were just full of merchandise, and during this process there's a change in mindset -- a paradigm shift -- so you have to involve merchandising for set improvements as soon as possible, to increase the variety of products you can put on the shelf to satisfy your customers. That's one of the benefits of all this. Before the pilot the shelves were stocked full. Once we started the pilot, there were a lot of holes. That's what will happen when you start this."
Moving forward, Gilmour plans to hire a perpetual inventory coordinator to help plan the training and rollout of the CAO system to the remaining stores. Then he'll improve the sets, assortment, and variety of product by using the additional 10 percent to 15 percent of space United is realizing from the system.
The SAF Superstore forecasting engine will eventually interface with the ACNielsen Spaceman shelf management system to develop planograms more efficiently.
Converting all of United's suppliers to detailed invoices is another crucial step that must be taken to move forward with the CAO implementation. "We want to expand all of our automated categories to include all dry grocery, frozen, dairy, and nonfoods," Gilmour says. "Right now we have a nonfoods supplier that doesn't provide us with detailed invoices, so we have to get that working with them. There's a real financial impact that can be achieved by including nonfoods, because you have a lot of product with high exposure and shrink, and a lot of expensive product in nonfoods."
Following this, United will reassess ROI to include the profitability of new assortments, and key measures such as sales per linear foot.
SKU-level support
To capture SKU-level support at the item level, Gilmour will install Tomax retail.net software. "This is really going to help with our perpetual inventory," he says. "You have repacks and different UPC codes, but you have the same SKU -- so we'll be able to keep up with 12-ounce cans of Coke that may be in different areas of the store. It will help us do a much better job of forecasting."
Finally, Gilmour plans to find a software provider to help United expand the CAO system into its perishable departments, to address in-store production and perishability aspects. "Hopefully, this will be another huge gain," Gilmour says. "As you know, when you deal with bakery, restaurant operations, seafood, meats, and cheeses, there's a tremendous opportunity to reduce shrink."
It's technologies such as this that will deliver real competitive advantages to United, according to Gilmour. "With the Internet generation on the rise, the cost of staying in business will be to stay abreast of the best technology that drives customer convenience, reduction in costs, and managed assortments for greater profitability," he says.
By any measure, United's CAO system will help the retailer deliver on its core philosophy: a commitment to customer service -- providing the best value for high-quality goods and treating each customer as he or she would want to be treated.
"Our c.e.o., Kent Moore, always says, 'We're in the customer service business; we just happen to make our money selling groceries,'" Gilmour says. "Anything that can help us to do that is extremely valuable."
Such was the case at United Supermarkets, Ltd., a 47-store chain based in Lubbock, Texas, which is in the midst of implementing a computer-automated ordering (CAO) system. "With any initiative like this, there are a lot of people in the operation who get a bit nervous," says United v.p. and c.i.o. Scott Gilmour, who spoke this past March at the FMI Marketechnics session called "Building a Knowledge-based Computer-Automated Ordering System." "Chief financial officers, for example, get nervous because you're changing to perpetual inventory from -- in our case -- manual accounting three times a year."
Automated ordering is just one of the many projects occurring under Gilmour's watch. United's technology infrastructure is divided into three areas: corporate systems, store systems, and supply chain/logistics systems, and Gilmour's staff is currently involved in various implementations in each of these, among them establishing an asset management system, integrating the fuel point of sale with the store point of sale, and building a Web-based labor-scheduling system.
Some of the pilot objectives Gilmour had for the CAO system were to reduce the number of out-of-stocks, automatically detect when an out-of-stock occurs, optimize the shelf stock by decreasing storage costs and inventory, reduce handling costs, and reduce ordering and inventory management labor.
Gilmour chose to use SofTechnics and SAF AG for the project. "We were the first grocer to install this technology from these companies," he says. "We wanted to help them to demonstrate and validate the solution, to evaluate the system's performance and measure the results, improve the capabilities, and identify product enhancements."
Vendor goals
SofTechnics and SAF, on the other hand, were looking to gain valuable project experience and to help ensure United's success during the technology companies' first venture into food retail. "We wanted this to be a win-win for both of us," Gilmour says. "We were also looking to achieve feasibility study key performance indicator results regarding out-of-stock reduction, and we wanted to reduce our inventory to optimal levels."
This wasn't going to be an easy task. The difference between computer-assisted ordering -- which is what United was doing -- and computer-automated ordering is that in the latter there's no human intervention. In computer-assisted ordering a suggested order is created by computer, and an employee is prompted to accept the order or make changes. In computer-automated ordering the computer creates and generates the order; it doesn't prompt someone for approval.
This means that if the store's inventory isn't accounted for, then the ordering system won't do its job. "There are a lot of disciplines needed for this," Gilmour points out. "Everything from 100 percent scanning to transfer from one department to another, for transfers to other stores, to reclamation, to throwaway, all those things have to be scanned. Obviously the detailed invoicing at the back end is the starting point, but there are a lot of disciplines that need to be put in place."
United needed to make some upgrades before it began its feasibility study in the summer of 2002. Gilmour's team had to upgrade its system from an earlier version of SofTechnics to a newer one and had to convert from DBfile to a more industrial-strength relational database, which was found in Informix.
Gilmour already had a jump on the second step, which was to create detailed invoicing. "We were already in production with detailed invoicing from our warehouse," he says. "The fact that we started with dry grocery items and frozen items -- for which we were already receiving detailed invoices -- was a major hurdle which we had already cleared."
Personnel issues
The next step was crucial and could make or break the project: getting the appropriate personnel into the two pilot stores and making sure they were trained to implement a process to scan all transfers and activate scan data collection.
The forecasting engine for the CAO system required two years of previous data for the feasibility study and pilot initiation. While United had detailed sales data for the previous years, it didn't have any promotional history. To address this problem, SAF developed a promotion detection script based on assumptions about sales spikes.
The last step was to count all of the product the evening before the system went live. "We had to physically count all of the items to initialize both the physical inventory quantities and to set up items for automated replenishment," Gilmour says. "There was some obvious labor involved with this. We needed an accurate count of what we had in the back rooms and on the shelves and on the displays for each item, and then we turned on the system after the count was completed."
Gilmour started with one store, then began the other pilot in a second store a week later, implementing the same process in each.
Looking back, Gilmour would have changed one item if he'd had the chance, and he recommends that any retailers undergoing such an implementation take note: "We decided upon major product lines, like paper products, canned fruits and canned vegetables, cereals, pet food, and frozen food," he says. "The one thing we didn't take into consideration was the opportunity to reset these areas. Later, after the initial pilot, we turned off a few of these products and did facing aisles so we could bring in our category management staff and employ our ACNielsen Spaceman shelf management system to do some resets."
Still, the results were encouraging. In one store, inventory was reduced 17 percent, and the other saw a 22 percent decrease. The value of the inventory was decreased 14 percent and 12 percent. Out-of-stocks dropped a whopping 58 percent in one store, and 62 percent in the other.
There was an increase in sales for each store undergoing the pilot, but Gilmour takes this with a grain of salt. "We can't say that this is just from the CAO system," he says. "There may have been a number of other things involved in our ability to increase sales."
Interestingly, the actual results were very close to the simulation that was performed prior to the start of the pilot.
Reduced labor
One of the goals of implementing the CAO system was to reduce the labor required for taking inventory. "We don't want to have somebody go around and physically count these items," Gilmour says. "We want the system to do it through the detailed invoices coming in the back door and through effective scanning of reclamations and transfers. This should be no problem if the system is tracking our inventory accurately."
During the pilot stages Gilmour found that only 6 percent of the items required manual intervention. "This was due to alerts that we set up in the system, such as too many zero sales on an item, a new listing, zero inventory, or a drop in sales," he explains. "An employee who's working in the store would count the item that was affected to keep the system accurate."
Any change to fully automated tasks can't be completed successfully without addressing issues of change management, and this was a topic Gilmour made sure he addressed upfront. "We wanted to empower our associates with an understanding of the benefits of the CAO system," he says. "The emphasis on the importance of inventory discipline was critical."
He also had to address the issue of resets. "When it came to resets, [associates] had to get used to the idea of facing rather than filling -- at least initially. Our stores were just full of merchandise, and during this process there's a change in mindset -- a paradigm shift -- so you have to involve merchandising for set improvements as soon as possible, to increase the variety of products you can put on the shelf to satisfy your customers. That's one of the benefits of all this. Before the pilot the shelves were stocked full. Once we started the pilot, there were a lot of holes. That's what will happen when you start this."
Moving forward, Gilmour plans to hire a perpetual inventory coordinator to help plan the training and rollout of the CAO system to the remaining stores. Then he'll improve the sets, assortment, and variety of product by using the additional 10 percent to 15 percent of space United is realizing from the system.
The SAF Superstore forecasting engine will eventually interface with the ACNielsen Spaceman shelf management system to develop planograms more efficiently.
Converting all of United's suppliers to detailed invoices is another crucial step that must be taken to move forward with the CAO implementation. "We want to expand all of our automated categories to include all dry grocery, frozen, dairy, and nonfoods," Gilmour says. "Right now we have a nonfoods supplier that doesn't provide us with detailed invoices, so we have to get that working with them. There's a real financial impact that can be achieved by including nonfoods, because you have a lot of product with high exposure and shrink, and a lot of expensive product in nonfoods."
Following this, United will reassess ROI to include the profitability of new assortments, and key measures such as sales per linear foot.
SKU-level support
To capture SKU-level support at the item level, Gilmour will install Tomax retail.net software. "This is really going to help with our perpetual inventory," he says. "You have repacks and different UPC codes, but you have the same SKU -- so we'll be able to keep up with 12-ounce cans of Coke that may be in different areas of the store. It will help us do a much better job of forecasting."
Finally, Gilmour plans to find a software provider to help United expand the CAO system into its perishable departments, to address in-store production and perishability aspects. "Hopefully, this will be another huge gain," Gilmour says. "As you know, when you deal with bakery, restaurant operations, seafood, meats, and cheeses, there's a tremendous opportunity to reduce shrink."
It's technologies such as this that will deliver real competitive advantages to United, according to Gilmour. "With the Internet generation on the rise, the cost of staying in business will be to stay abreast of the best technology that drives customer convenience, reduction in costs, and managed assortments for greater profitability," he says.
By any measure, United's CAO system will help the retailer deliver on its core philosophy: a commitment to customer service -- providing the best value for high-quality goods and treating each customer as he or she would want to be treated.
"Our c.e.o., Kent Moore, always says, 'We're in the customer service business; we just happen to make our money selling groceries,'" Gilmour says. "Anything that can help us to do that is extremely valuable."