Any American company stuck in the early stages of a Collaborative Planning, Forecasting and Replenishment (CPFR) pilot would do well to look across the ocean for guidance. In Spain, a major provider of cleaning products has realized dramatic results with one of the country's largest grocery chains.

The partners are Henkel KGaA, the Dusseldorf, Germany-based maker of household cleaners, adhesives, toiletries and other home care products, and Grupo Eroski, the Spanish grocery chain whose thousands of outlets range from "hypermarkets" to mini-markets. Together they launched a successful CPFR pilot, at a time when few such efforts were in evidence in Europe or the U.S.

Henkel may not be a household name in the U.S., but it's tough to ignore in Europe, which accounts for 50 percent of sales. The diversified company has sold off or acquired numerous entities in its 126 years of existence. In 2001, it shed two chemical-making units, Cognis and Henkel-Ecolab. The move reflected a desire to focus on its core product line of detergents, cleansers, adhesives and toiletries. Henkel's current U.S. holdings include Manco, the maker of Duck Tape, and a 27-percent share in The Clorox Co.

	It took about a year for Henkel to get the CPFR program's elements fully in place.

At a time when many consumer-goods manufacturers are struggling, Henkel has done reasonably well. It showed an operating profit of 168m euros in the second quarter of 2002, up 11.5 percent over the same period of the previous year. In the first six months of 2002, sales of the Henkel Group rose 3.3 percent, to 4.9bn euros. Still controlled by the Henkel family, the company has about 47,000 employees around the world, doing business in 70 countries.

Beneath the surface, though, Henkel's operations haven't been so consistent. In the late 1990s, the company was suffering from serious flaws in its forecasting and execution methods. Inventory levels were unacceptably high, yet stockouts were chronic. So were transportation inefficiencies.

Much of the problem stemmed from an inability to gauge the amount of product needed in a given market. According to Esteban Garriga, trade services manager of Henkel Iberica in Barcelona, the company was experiencing sales-forecast accuracy of between 40 and 45 percent. Clearly there was a need for better ties with retailers.

But Henkel didn't rush blindly into a collaboration project. First, says Garriga, it needed to get its own house in order. That meant implementing a new demand-planning system for internal forecasting. At the time, the company had no coherent system, was unable to integrate a continuous replenishment program into the forecast, and couldn't even communicate the basic events on which it relied. Forecasts were produced through Excel spreadsheets. The results were no surprise: high costs and poor service.

Henkel's goals were equally clear. It wanted to eliminate a slew of internal glitches, including stockouts, delivery errors and invoicing complaints. Forecasts had to be synchronized with production at one end, and consumer demand at the other. Purchasing and distribution, two of the corporate world's most notorious "silos," had to be integrated. Sales and marketing had to be kept in the loop as well, for access to key demand data.

The answer, at least from an internal standpoint, lay in acquisition of the Demand Planning (DP) module of the Rockville, Md.-based Manugistics Group, Inc. The purchase was accompanied by a wholesale reengineering of Henkel's business processes.

Manugistics' DP tool provides 95 percent of the company's total forecast. It draws on information from marketing and sales, among other sources. At the same time, Henkel took pains to define key performance indicators (KPIs), mostly related to forecast accuracy and promotions planning. (Promotions were among the areas giving Henkel the most trouble, Garriga says.)

Henkel's fulfillment analysis incorporated such key elements as the planning horizon, modifications to the plan, associated promotional volumes, and internal follow-up by various players in the organization. From that analysis, the company derived an event plan that identified who was responsible for each step of the process. Internal forecasts, especially those related to special programs or promotions, became a joint effort for the first time.

Revised Forecasts
They also became considerably more accurate. Henkel went from bi-monthly forecast revisions to weekly and even daily adjustments. What's more, the forecasts were based on a far larger universe of data than before, supplied by all departments of the company. And the resources need to process the data declined sharply. Before, says Henkel, staff would spend 40 percent of its time on data analysis and management, and the rest on processing. Following implementation of the DP and demand-forecasting tools, the same individuals were spending 80 percent of their time on analysis.

Implementation of the DP software was just the first step in Henkel's quest for forecasting accuracy. "Even with Manugistics, it was not enough," says Garriga. "In the end, it doesn't help if the information you put into the planning systems is wrong." And promotions can still go askew if the manufacturer can't react to sudden changes in demand.

Forecasting priorities of manufacturer and retailer differ in key areas, says Garriga. The former focuses on shipments, order-lead time, production capacity, product availability, promotions and raw-materials supply. The latter looks at visibility to consumer demand, in-stock position, variance in shipments, promotional activity, growth plans, and distributor structure.

Somehow those disparate concerns must be melded into a coherent forecasting model. For Henkel, the answer lay in CPFR, the set of collaborative processes which has attracted heavy attention in the U.S. and elsewhere, but has so far yielded little in the way of completed projects. That was even truer back in 1999, when Henkel, its DP software in place, began casting about for a CPFR partner.

It didn't look for long. Eroski, says Garriga, was Henkel's "customer number one" in Spain. The two were doing a significant amount of business together, but the processes underlying that partnership left much to be desired. Half of Henkel's sales forecasts had an average error of more than 50 percent. Stockouts were common, especially on promotional items. Shipments would frequently arrive late at Eroski's central warehouse, which was servicing 500 stores.

One of Spain's leading grocery retailers, Eroski was a natural candidate for the experiment. Focused mainly on the Basque region, the company operates 47 hypermarkets, 800 supermarkets under the Consum brand, and some 2,000 Charter mini-markets. It also has three hypermarkets and 17 supermarkets in France. In addition to the central warehouse, Eroski maintains a number of regional distribution platforms, along with direct-store delivery for distant points or selected products.

In crafting a CPFR program, the partners had little in the way of examples on which to draw. At the time, says Garriga, there were no mature projects in Europe, other than a couple of isolated efforts in Italy. Not much had been completed in the U.S. either, although Henkel drew heavily on groundwork laid by the Voluntary Inter-industry Commerce Standards Association (VICS), and the experience of CPFR pioneers such as former Nabisco executive Joseph Andraski.

Work began in late 1999. Henkel and Eroski were aided by Accenture (then Andersen Consulting), which acted as consultant and systems integrator. Accenture had already defined the CPFR standard for processes and information exchange on behalf of two dozen companies (including Henkel) that make up the group known as ECR Europe. (ECR, for Efficient Consumer Response, is a collaborative planning effort within the grocery industry, which predates CPFR.) That effort led directly to Henkel's individual pilot.

On the software side, Henkel once again chose Manugistics, whose NetWORKS Collaborate module provided the platform for information-sharing. The customer was the first in Europe to implement Manugistics' collaboration product, according to Martine Gosse, the vendor's Paris-based marketing director for Europe. The pilot involved the entire category of detergent products sold by Henkel through Eroski's markets.

Opening the Pipeline
Its own internal forecasting processes finally in place, Henkel was able to assume sole responsibility for providing the production forecast, figuring proper supply levels and generating orders. Yet it still needed a free-flowing pipeline for data between itself and Eroski. The partners began exchanging information once a day on outgoing stock, stock figures and orders; once a week on order forecasts; every 15 days on sales forecasts (eventually through the NetWORKS Collaborate tool), and every four months on the promotional events calendar.

With the help of the internet, Henkel and Eroski developed common business and promotional plans, compared sales forecasts, order forecasts and exceptions, and obtained information on changes in promotions and product availability relative to demand, among other things.

They also paid close attention to measurements. The pilot was built around a select number of KPIs, including customer-service levels at Eroski's central warehouse, number of stockouts, number of promotions, stock rotation, forecast reliability, truck fill rates, pallet fill rates, and number of urgent orders.

No major process change is without obstacles. Jaume Ferrer, Accenture's partner responsible for supply chain in continental Europe, says the major one was human in nature. At the pilot stage, he says, CPFR is largely a matter of change management. Entrenched practices, such as manufacturers offering discounts to stimulate sales long before goods are actually bought by consumers, must be stopped.

Gosse agrees that a "mentality change" was necessary in order for the partners to collaborate fully. CPFR is still a relatively new concept in Europe, she says. So is the notion of supply-chain management, whereby companies tear down barriers between traditional functions and encourage the free flow of data.

With the CPFR pilot in place, the quality of Henkel's sales forecasts improved steadily. In the period between October 1999 and March 2000, the share of forecasts showing an average error of more than 50 percent declined from nearly half of the total to around 5 percent. Meanwhile, forecasts with an error rate of less than 20 percent - a reasonable level, given the vagaries of consumer taste - grew from 20 percent to 75 percent.

Other KPIs during that period showed similarly strong results: a 98-percent customer-service level, five days of supply, 2-percent stockouts, better than 85-percent forecast reliability, 98-percent truck fill rates, and 99-percent pallet fills. Perhaps most importantly, says Garriga, was the integration of data on new-product introductions with that concerning local activities, into a single, reliable forecast. Previously, such information was not a regular part of logistics plans.

The partners set up a relatively small team to get the pilot rolling. Henkel contributed five people, including a consultant from Manugistics, while Eroski supplied four. They were backed by internal information technology staffers, as well as a team of Accenture consultants. The latter were instrumental in plotting how the pilot could be expanded to include additional products, Garriga says.

Of special note on the personnel side was the companies' insistence on the involvement of sales and customer-service staff. Sales people have a poor record of participation in such partnerships, says Garriga. And customer-service reps aren't schooled in the intricacies of sales forecasting or commercial planning. But the market knowledge of those individuals is too valuable to waste. So Henkel's key account manager for Eroski, not a demand-planning expert, was chosen to spearhead collaboration that would lead to creation of a sales forecast. And the customer-service team was given responsibility for incorporating promotions into the forecast.

Demand, Production Linked
The demand planner, meanwhile, focused on synchronizing that end of the chain with production planning. To avoid supply glitches, plans would be squared with capacity constraints before figures were transmitted to Eroski. In the event of a discrepancy, the system would generate a warning message. For the pilot, Henkel established a forecasting horizon of five weeks, giving production enough time to get its operations in line. The period would later be expanded to two months.

Within six months of launching the CPFR pilot, Henkel presented its preliminary results at an ECR conference in Europe. But it took another six months, says Garriga, to get all of the program's elements into place.

Along the way, the partners discovered how hard it was to implement all nine steps of the CPFR process model as developed by VICS. They can be broken down into three categories - planning, forecasting and replenishment - with the final one focusing on collaborative order generation.

Henkel did end up addressing most of the nine steps, but only in the form of a high-level business case, aided by Accenture, on the feasibility of implementing CPFR throughout Europe. And, in the case of the Eroski project, it bypassed at least one step - the use of technology tools for validating exceptions - altogether. Instead, the company is relying on the telephone and e-mail for that purpose.

CPFR is still a relatively new concept in Europe. So is supply-chain management, which encourages the free flow of data.

A key lesson, says Garriga, was that each customer requires a unique focus, emphasizing some stages at the expense of others. The Eroski pilot centered on collaborative forecasting, an area in which Henkel had considered itself especially weak. A subsequent CPFR effort, with the Catalonian grocery retailer Condis, tackled up-front agreements leading to promotional calendars. According to Garriga, the software tools created to support CPFR might focus on one area or the other, but never both.

It's easy to get lost in the mass of requirements that make up the CPFR model. Garriga recommends that companies avoid treating the effort as a single "super project." Rather they should approach it as a series of 20 to 30 smaller, more manageable tasks. Such an approach can also be helpful in implementing CPFR with smaller customers.

Henkel also learned about the importance of maintaining a CPFR process indefinitely. Long after a pilot has been implemented, companies must keep in place a task force to oversee its continued success, Garriga says.

Henkel's success with Eroski led directly to the subsequent pilot with Condis, which operates 300 stores in Spain's Catalonia region under the brand names of Condis Supermarkets and Distop. This time around, Henkel sought to include not just the manufacturer and retailer, but an upstream supplier. The goal, says Garriga, was end-to-end integration of the supply chain through the deployment of CPFR.

Launched in September 2000, the Condis pilot was completed earlier this year. The relatively low-tech effort involved the use of spreadsheets for data, with collaboration taking place through e-mail and telephone. The focus this time was on promotional SKUs.

Jointly defined KPIs in the Condis pilot included forecasting accuracy, promotion planning, promotion plan changes, service levels, inventory levels, number of rush orders, and the use of full pallets and full trucks. Initial results showed a 15-percent improvement in forecast accuracy between January and July of 2001, along with better joint planning of promotions. Customer- service levels climbed to 99 percent without an increase in inventory levels. Overall supply-chain costs fell 6 percent, largely due to a decline in rush orders.

From the start, says Garriga, Henkel's biggest challenge was getting its arms around the complex CPFR model, and what was needed to apply it. "We had to know all of the drivers of the forecast," he says. "Then we had to decide what to focus on."

Longer term, CPFR continues to be impeded by a general reluctance by supply-chain partners to collaborate in a meaningful way. Many tout the virtues of collaboration, but few seem willing to undertake the revamping of business processes, and sharing of sensitive information, that make it possible.

Further blocking progress is a continued lack of communications standards for global CPFR, a deficiency that members of VICS, the ECR movement and the Global Commerce Initiative, among others, are working to rectify. Still, that hasn't stopped some 50 European companies from moving beyond the pilot stage to full use of CPFR, says Ferrer. And Gosse points to the work of GlobalNetXchange, L.L.C. (GNX), an electronic marketplace which includes some of the largest retailers in Europe and the U.S. GNX features a supply-chain collaboration suite built around the CPFR model.

As for Henkel, it is bent on extending its success in the Eroski and Condis pilots to other customers during the coming year. Following its own advice, the company will take modest steps toward that goal. "Instead of big projects," says Garriga, "we'll be looking for more smaller ones that we can easily implement."