Executive Briefings

Special Issue: Collaborative Commerce - Forecasting Tool Lowers Coke Bottler's Inventory

Enabling its field sales managers to collaborate on forecasts allowed Coca-Cola Bottling Co. Consolidated to slash inventories in half while absorbing 150 new products.

When Coca-Cola came only in little green bottles and was the soft-drink maker's single brand, replenishing stores was a simple matter. Drivers loaded their trucks with product and dropped whatever was needed at the businesses on their route.

The environment at Coke's bottlers and distributors today is far more complex, as evidenced by Coca-Cola Bottling Co. Consolidated in Charlotte, N.C. CCBCC is Coca-Cola's second-largest bottler in the U.S. and one of the highest per-capita soft drink bottlers in the world. It bottles, packages and distributes 13,000 SKUs to 200,000 customers throughout 11 Southeastern states, delivering more than 125 million cases of product each year. Nearly all of its customers engage in frequent promotions and aggressive discounting - often with little or no notice to the bottler - that can result in dramatic swings in volume for particular products. With five bottling plants and 60 distribution centers, knowing what to produce and how much to ship to which locations are critical issues.

Before implementing a collaborative demand planning and forecasting system from Manugistics, Rockville, Md., two years ago, these issues were "pain points" for the bottling company. "We were very efficient at producing product, but we really didn't know what we should be making," says Brian Wieland, director of demand planning at CCBCC.

It wanted a collaborative solution on the front end to enable its field sales staff to have real-time, dynamic input into the forecasting process.

"We simply weren't able to react quickly enough to our market," he continues. "We are in a very retail-driven sales environment, and on any given day one of our major retail chains might alter what they are promoting in their next flyer. For example, a grocery store chain might decide to feature a two-litre bottle of Classic Coke as opposed to a 12-pack. That can change the whole dynamic of what we need to produce and ship. By the time we realized we were not producing the right product, we already had a lot of the wrong inventory sitting on the floor."

The result was that the company suffered sporadic stockouts while holding too much inventory overall. Additionally, as the company continued to grow and as Coca-Cola introduced more new products, CCBCC was facing the expense of adding 10 to 15 new warehouses to its then already large network of 73. Since warehouses average $5m each, this represented a huge capital investment.

When the company began looking for solutions, it basically had three goals in mind, says Wieland. First, it wanted to cut finished-goods inventory in half, from 12-14 days to six - four days at the plant and two days at the distribution center, or in the store for the small number of instances where CCBCC ships direct. This achievement also would result in fresher product on the shelf. Second, it wanted to improve customer service levels and reduce stockouts by having the right product in the right place. And, finally, it wanted to reduce large-scale capital investments in warehousing.

Demand Visibility
To accomplish these goals, the company knew it needed to have better visibility to actual demand and more flexibility in production. It wanted a collaborative solution on the front end, not to get customers directly involved - a difficult step with its huge and diverse customer base - but to enable its field sales staff to have real-time, dynamic input into the forecasting process.

This approach is fairly typical of early adopters implementing a collaborative solution, according to Nick Asonsky, director of Manugistics' solutions consulting group. "I would guess that half of the clients who have bought our collaborative tool are using it for sales and operations planning or for internal forecast collaboration," he says. "They may end up doing collaborative planning, forecasting and replenishing with customer involvement later on, but internal collaboration usually is the first goal."

CCBCC decided to partner with Manugistics after reviewing several possible vendors. "We felt that Manugistics had the most well-rounded solution," says Wieland. The company elected to implement Networks Demand, Networks Fulfillment, Networks Collaborate, Production Scheduling and Demand Planning Extended Edition. The demand planning and scheduling came first, and then the collaboration piece was added. Manugistics' Transportation Management module will be installed in the next couple of months.

The project drove a fundamental change in the way CCBCC managed its business. Previously, forecasting and production had been completely decentralized, says Wieland. Each manufacturing plant created its own forecasts and scheduled production and delivery to its associated DCs based on that. Forecasts typically looked out four weeks and were updated weekly by that plant's sales and marketing personnel. There was no visibility between plants or distribution centers, so an inventory surplus in one location could not easily be shifted to make up for a shortage elsewhere.

Centralized Planning
Under the new system, forecasting, production and replenishment are managed by a centralized planning group based at the Charlotte headquarters. "It's a very interactive atmosphere and allows for a lot of cross-functional activity, with people working in all three areas," says Wieland. With centralized management, the company has visibility throughout all its facilities. And because the entire forecast and production plan can be rerun in less than an hour, "we are able to be far more proactive than before, instead of merely reactive," Wieland says.

"With the Collaborate application, our sales managers are able to provide local intelligence to the system."
- Brian Wieland of CCBCC

The current process begins with the marketing and customer-development group feeding in price plans from the company's largest customers - "those that in a given week can run us out of stock if they have a big ad," says Wieland. Estimated demand from other customers, along with known promotions and price points, also is input, as is information about new product introductions or other systemwide events. Wieland notes that Coca-Cola has added about 150 new items in the past year, including Vanilla Coke, Fanta drinks and Dasani water products. "Under the old system," he says, "it would have been impossible to plan all those new items and to manage all the required changeovers to our production lines."

Since large chains give CCBCC's representatives their promotional plans on a trimester basis, the company's base forecast presents a 17-week forward view. Attention is focused on the next four weeks, however. Every Thursday the forecast is updated and sent to the area sales managers, who are aligned with specific distribution centers. They review the forecast, making notes and modifications as far as four weeks out. These modifications are based on new or changing promotions among their customer base and on specific local knowledge, such as a fair coming to town or foul weather that may prompt people to stock up on bottled water.

"With the Collaborate application, our sales managers are able to provide local intelligence to the system, and they also can see their forecast accuracy in the past as well as any adjustments they already have made," says Wieland.

Reality Check
This input is not a blank check. If changes alter the initial forecast more than a set amount, the system triggers an automatic alert to the demand analyst, who can communicate with that sales manager. "This is just an additional reality check," says Wieland. "Sometimes you just get somebody keying in a wrong number, and sometimes a sales manager might be a little overly optimistic about what is going to sell. This is a good double-check."

On days subsequent to the Thursday update, sales managers also can send messages to demand analysts and go in and make changes to the forecast based on more current information. "We want to get the information in there as soon as possible because we rerun the plan every night to bring in that day's changes in inventory positions," says Wieland. These changes reflect inventory that was shipped from the DCs on that day as well as what was received from the plants, but it does not incorporate point-of-sale information. "It would be nice to have POS data from our biggest customers, but we have a lot of small customers who just don't have the capability," says Wieland, "so we just go off the deliveries."

The demand plan drives production and scheduling, which is locked in three days out. Distribution is planned every day, with shipments going out in the evening for next-day delivery.

When Manugistics' Transportation Management module is added in a month or so, load building will become more efficient says Wieland. "This solution will take all the orders for that day and combine them in an optimal way into truckload shipments. It will prioritize those loads a little better."

The transportation solution also will help CCBCC communicate and collaborate with its common carriers, which are used in addition to a private fleet. "We will be able to tender our loads out to common carriers electronically," says Wieland. The result will be better utilization of lane agreements and a carrier selection process that includes data on past performance, he says. In addition, a freight payment system will be used to pay carriers in a paperless manner. "The last piece will be driver dispatching," Wieland says. "We will be able to better schedule drivers and take advantage of backhaul opportunities. Right now that is pretty much done using Excel spreadsheets - it's really a manual process."

Goals Met
Already the benefits from this project are significant and have met or exceeded CCBCC's initial goals, Wieland reports. Overall inventory has been reduced by about half. Even that figure underestimates the impact, Wieland says, because in the same time period that those reductions were realized at least 150 new stock-carrying units were added. "We deliver in pallet quantities and each of those takes up a footprint," he says. "So when you are introducing new products that add inventory and are still able to reduce inventory overall by a significant amount, that is quite remarkable."

The inventory reductions occurred very rapidly once the system was in place, he says. "You get that quick hit and then you keep working to gradually bring it down more as you tweak the system and learn how to use it better."

"Given the dynamic marketplace we operate in, being able to react on the fly like that is really an important benefit."
- CCBCC's Wieland

Forecast accuracy has improved 10 percent to 20 percent and now is above 90 percent accurate, Wieland says. "What is really so important about this is that we can change the forecast as things happen, and within an hour or less we can rerun the entire plan and know what our new needs are - what we now need to replenish, what we need to schedule from a production standpoint, and even the raw materials we now need to order. In matter of minutes we can have that whole new look. Given the dynamic marketplace we operate in, being able to react on the fly like that is really an important benefit."

Finally, the company has achieved significant reductions in capital expenditure. "This was a really big one," says Wieland. Not only did CCBCC refrain from spending approximately $50m on new warehouses, it actually closed more than a dozen existing facilities - a process that is ongoing. "We cut the inventory so much that we get a lot more efficient use of our warehouse space," says Wieland. "Before it seemed that once we filled up a warehouse we had no choice but to build a new one."

Looking to the future, CCBCC anticipates doing a lot more collaboration with its suppliers. "We want to start feeding our suppliers information on a regular basis because when our plans change, their plans have to change," says Wieland. "When our need for can-bodies goes up, they had better be available or we are out of luck."

Wieland says CCBCC already shares information on a weekly basis with a number of its largest suppliers. "But we would like to take this a step further and allow them access to our system on an ad hoc basis so that they can go in and pull up their numbers in the format they need. What we would like to do is move to vendor-managed inventory, where our suppliers could manage our inventory and create all our moves of raw materials."

When Coca-Cola came only in little green bottles and was the soft-drink maker's single brand, replenishing stores was a simple matter. Drivers loaded their trucks with product and dropped whatever was needed at the businesses on their route.

The environment at Coke's bottlers and distributors today is far more complex, as evidenced by Coca-Cola Bottling Co. Consolidated in Charlotte, N.C. CCBCC is Coca-Cola's second-largest bottler in the U.S. and one of the highest per-capita soft drink bottlers in the world. It bottles, packages and distributes 13,000 SKUs to 200,000 customers throughout 11 Southeastern states, delivering more than 125 million cases of product each year. Nearly all of its customers engage in frequent promotions and aggressive discounting - often with little or no notice to the bottler - that can result in dramatic swings in volume for particular products. With five bottling plants and 60 distribution centers, knowing what to produce and how much to ship to which locations are critical issues.

Before implementing a collaborative demand planning and forecasting system from Manugistics, Rockville, Md., two years ago, these issues were "pain points" for the bottling company. "We were very efficient at producing product, but we really didn't know what we should be making," says Brian Wieland, director of demand planning at CCBCC.

It wanted a collaborative solution on the front end to enable its field sales staff to have real-time, dynamic input into the forecasting process.

"We simply weren't able to react quickly enough to our market," he continues. "We are in a very retail-driven sales environment, and on any given day one of our major retail chains might alter what they are promoting in their next flyer. For example, a grocery store chain might decide to feature a two-litre bottle of Classic Coke as opposed to a 12-pack. That can change the whole dynamic of what we need to produce and ship. By the time we realized we were not producing the right product, we already had a lot of the wrong inventory sitting on the floor."

The result was that the company suffered sporadic stockouts while holding too much inventory overall. Additionally, as the company continued to grow and as Coca-Cola introduced more new products, CCBCC was facing the expense of adding 10 to 15 new warehouses to its then already large network of 73. Since warehouses average $5m each, this represented a huge capital investment.

When the company began looking for solutions, it basically had three goals in mind, says Wieland. First, it wanted to cut finished-goods inventory in half, from 12-14 days to six - four days at the plant and two days at the distribution center, or in the store for the small number of instances where CCBCC ships direct. This achievement also would result in fresher product on the shelf. Second, it wanted to improve customer service levels and reduce stockouts by having the right product in the right place. And, finally, it wanted to reduce large-scale capital investments in warehousing.

Demand Visibility
To accomplish these goals, the company knew it needed to have better visibility to actual demand and more flexibility in production. It wanted a collaborative solution on the front end, not to get customers directly involved - a difficult step with its huge and diverse customer base - but to enable its field sales staff to have real-time, dynamic input into the forecasting process.

This approach is fairly typical of early adopters implementing a collaborative solution, according to Nick Asonsky, director of Manugistics' solutions consulting group. "I would guess that half of the clients who have bought our collaborative tool are using it for sales and operations planning or for internal forecast collaboration," he says. "They may end up doing collaborative planning, forecasting and replenishing with customer involvement later on, but internal collaboration usually is the first goal."

CCBCC decided to partner with Manugistics after reviewing several possible vendors. "We felt that Manugistics had the most well-rounded solution," says Wieland. The company elected to implement Networks Demand, Networks Fulfillment, Networks Collaborate, Production Scheduling and Demand Planning Extended Edition. The demand planning and scheduling came first, and then the collaboration piece was added. Manugistics' Transportation Management module will be installed in the next couple of months.

The project drove a fundamental change in the way CCBCC managed its business. Previously, forecasting and production had been completely decentralized, says Wieland. Each manufacturing plant created its own forecasts and scheduled production and delivery to its associated DCs based on that. Forecasts typically looked out four weeks and were updated weekly by that plant's sales and marketing personnel. There was no visibility between plants or distribution centers, so an inventory surplus in one location could not easily be shifted to make up for a shortage elsewhere.

Centralized Planning
Under the new system, forecasting, production and replenishment are managed by a centralized planning group based at the Charlotte headquarters. "It's a very interactive atmosphere and allows for a lot of cross-functional activity, with people working in all three areas," says Wieland. With centralized management, the company has visibility throughout all its facilities. And because the entire forecast and production plan can be rerun in less than an hour, "we are able to be far more proactive than before, instead of merely reactive," Wieland says.

"With the Collaborate application, our sales managers are able to provide local intelligence to the system."
- Brian Wieland of CCBCC

The current process begins with the marketing and customer-development group feeding in price plans from the company's largest customers - "those that in a given week can run us out of stock if they have a big ad," says Wieland. Estimated demand from other customers, along with known promotions and price points, also is input, as is information about new product introductions or other systemwide events. Wieland notes that Coca-Cola has added about 150 new items in the past year, including Vanilla Coke, Fanta drinks and Dasani water products. "Under the old system," he says, "it would have been impossible to plan all those new items and to manage all the required changeovers to our production lines."

Since large chains give CCBCC's representatives their promotional plans on a trimester basis, the company's base forecast presents a 17-week forward view. Attention is focused on the next four weeks, however. Every Thursday the forecast is updated and sent to the area sales managers, who are aligned with specific distribution centers. They review the forecast, making notes and modifications as far as four weeks out. These modifications are based on new or changing promotions among their customer base and on specific local knowledge, such as a fair coming to town or foul weather that may prompt people to stock up on bottled water.

"With the Collaborate application, our sales managers are able to provide local intelligence to the system, and they also can see their forecast accuracy in the past as well as any adjustments they already have made," says Wieland.

Reality Check
This input is not a blank check. If changes alter the initial forecast more than a set amount, the system triggers an automatic alert to the demand analyst, who can communicate with that sales manager. "This is just an additional reality check," says Wieland. "Sometimes you just get somebody keying in a wrong number, and sometimes a sales manager might be a little overly optimistic about what is going to sell. This is a good double-check."

On days subsequent to the Thursday update, sales managers also can send messages to demand analysts and go in and make changes to the forecast based on more current information. "We want to get the information in there as soon as possible because we rerun the plan every night to bring in that day's changes in inventory positions," says Wieland. These changes reflect inventory that was shipped from the DCs on that day as well as what was received from the plants, but it does not incorporate point-of-sale information. "It would be nice to have POS data from our biggest customers, but we have a lot of small customers who just don't have the capability," says Wieland, "so we just go off the deliveries."

The demand plan drives production and scheduling, which is locked in three days out. Distribution is planned every day, with shipments going out in the evening for next-day delivery.

When Manugistics' Transportation Management module is added in a month or so, load building will become more efficient says Wieland. "This solution will take all the orders for that day and combine them in an optimal way into truckload shipments. It will prioritize those loads a little better."

The transportation solution also will help CCBCC communicate and collaborate with its common carriers, which are used in addition to a private fleet. "We will be able to tender our loads out to common carriers electronically," says Wieland. The result will be better utilization of lane agreements and a carrier selection process that includes data on past performance, he says. In addition, a freight payment system will be used to pay carriers in a paperless manner. "The last piece will be driver dispatching," Wieland says. "We will be able to better schedule drivers and take advantage of backhaul opportunities. Right now that is pretty much done using Excel spreadsheets - it's really a manual process."

Goals Met
Already the benefits from this project are significant and have met or exceeded CCBCC's initial goals, Wieland reports. Overall inventory has been reduced by about half. Even that figure underestimates the impact, Wieland says, because in the same time period that those reductions were realized at least 150 new stock-carrying units were added. "We deliver in pallet quantities and each of those takes up a footprint," he says. "So when you are introducing new products that add inventory and are still able to reduce inventory overall by a significant amount, that is quite remarkable."

The inventory reductions occurred very rapidly once the system was in place, he says. "You get that quick hit and then you keep working to gradually bring it down more as you tweak the system and learn how to use it better."

"Given the dynamic marketplace we operate in, being able to react on the fly like that is really an important benefit."
- CCBCC's Wieland

Forecast accuracy has improved 10 percent to 20 percent and now is above 90 percent accurate, Wieland says. "What is really so important about this is that we can change the forecast as things happen, and within an hour or less we can rerun the entire plan and know what our new needs are - what we now need to replenish, what we need to schedule from a production standpoint, and even the raw materials we now need to order. In matter of minutes we can have that whole new look. Given the dynamic marketplace we operate in, being able to react on the fly like that is really an important benefit."

Finally, the company has achieved significant reductions in capital expenditure. "This was a really big one," says Wieland. Not only did CCBCC refrain from spending approximately $50m on new warehouses, it actually closed more than a dozen existing facilities - a process that is ongoing. "We cut the inventory so much that we get a lot more efficient use of our warehouse space," says Wieland. "Before it seemed that once we filled up a warehouse we had no choice but to build a new one."

Looking to the future, CCBCC anticipates doing a lot more collaboration with its suppliers. "We want to start feeding our suppliers information on a regular basis because when our plans change, their plans have to change," says Wieland. "When our need for can-bodies goes up, they had better be available or we are out of luck."

Wieland says CCBCC already shares information on a weekly basis with a number of its largest suppliers. "But we would like to take this a step further and allow them access to our system on an ad hoc basis so that they can go in and pull up their numbers in the format they need. What we would like to do is move to vendor-managed inventory, where our suppliers could manage our inventory and create all our moves of raw materials."