Forecasting has never been more important--or harder. Customers are less loyal, and global competition more fierce, making it difficult to predict where sales are going. Adding to the problem: Products, sales and distribution channels all have proliferated, and the life spans of products have gotten shorter.

As a result, some companies are being forced to adopt new ways to improve forecasting and planning. And a common theme links them all: collaboration. More specifically, these companies are requiring different departments--chiefly sales, production and marketing--to share more information and work together on setting sales and production goals. They are regularly reviewing how close forecasts come to actual results, and making adjustments in production and marketing as needed. They also are increasingly using full-time demand and supply planners who prepare forecasts and related recommendations about demand and production.

Cutting-edge companies take collaboration further, integrating operations with vendors and suppliers in ways that give each party access to data that helps keep the supply chain flowing and inventories lean. Once such links are established, a manufacturer, for example, no longer has to guess at a vendor's inventory or future promotional plans, hence forecasts--and sales--improve.

We expect to see widespread use of such processes someday. For now, though, many companies pursue collaborative forecasting in ways that limit its effectiveness, if they pursue it at all. In some cases, not enough departments are involved, while in others, planning meetings aren't held regularly, or support from management is lukewarm at best.

What follows, based on our research and experience, are seven rules companies can follow to make the most of collaboration in their forecasting efforts.

1. Get Senior Executives Involved: Strong forecasting needs support from senior executives because of the resources required. For starters, executives need to approve the purchase and installation of state-of-the-art technology that will increase and enhance collection and sharing of data. They also need to hire professionals trained in the latest methods of planning and forecasting. And to increase cooperation between departments, a change in corporate culture may be necessary. None of these things tend to happen without the commitment of senior executives.

One way to get the attention of key executives is to calculate what a one-percentage-point improvement in forecast accuracy may mean to the company. As supplies come closer to demand, customers can buy more, stores return less, and more revenue goes straight to the bottom line instead of paying for excess storage and handling. For a large company, it could add millions of dollars to the bottom line.

A look at Johnson & Johnson Co.'s LifeScan unit, where one of the co-authors of this article formerly worked, shows the difference-committed executives can make. In 1999, sales forecasts for the maker of diabetic-testing materials and equipment were driven by the sales and marketing departments using standard business software. The unit-volume forecasts proved overoptimistic for several consecutive periods, and large amounts of inventory piled up. Senior J&J managers then decided to step in, implementing a more collaborative, cross-functional process. They designed a task force to fully implement a planning process using monthly sales and operations meetings. This required senior marketing managers to attend supply-planning review meetings to better understand demand and supply issues. Marketing directors were required to review summaries of those meetings so they could speak effectively on key issues in monthly meetings with senior executives.

The change effectively raised the level of responsibility for aligning marketing and supply outlooks to the level of senior managers and directors, requiring each of them to sign off on the monthly sales, marketing and supply plans. With that, results improved, and today, such procedures and state-of-the-art forecasting tools are used throughout J&J. A spokesman for the New Brunswick, N.J., company declined to comment.

2. Explain the Mutual Benefits: Some parties will resist being asked to play a bigger role in forecasting unless they know what's in it for them.

Salespeople, for example, want to focus on selling, not forecasting. But if they understand that a more efficient supply chain helps make a product available when the customer wants it, thereby increasing sales commissions, they get interested. For months, Sean Reese, a demand planner at Ocean Spray Cranberries Inc., Lakeville-Middleboro, Mass., says he sought input from salespeople in order to make his sales forecasts more accurate. Finally, he explained to the salespeople's manager how better forecasts could help their sales by improving the supply chain and reducing instances of stores running out of stock--an event that, if repeated too many times, can lead stores to switch suppliers. Then the data started flowing in. Mr. Reese left Ocean Spray in 2005 and now works as a senior software developer and engineer for Fidelity Investments.

Arasco, a Saudi-based company with a core business of animal feeds, promised price cuts to distributors if they would take part in joint forecasts; more accurate forecasts, the company surmised, would help it ship products to the distributors immediately as the products came off the line, reducing Arasco's warehousing needs and costs, and helping the distributors maintain steady supplies. In a pilot program started in 2006, the company says, it promised participating distributors that it would share savings that resulted by cutting their prices by 2% to 4%. The company says the program was a great success. Forecast error fell to 9% from 15%, on-time deliveries rose to 93% from 85%, and the company expanded the program to other distributors and to customers of raw materials.

3. Clearly Define Goals and Agreements: The most obvious way to see that a company's forecasting is improving is when its supply chain becomes more efficient. Setting clear goals and metrics, such as reducing the number of days of inventory on hand, are musts. Cincinnati-based Proctor & Gamble Co. uses a scorecard that looks at on-time deliveries and the number of times a store runs out of a product, among other things.

Goals should be determined with input from all of a company's functions--since one department may have unrealistic expectations of another--as well as from suppliers and vendors. Review of metrics in regular meetings is essential to see whether goals are met, and where improvements are needed.

Collaboration with third parties requires detailed agreements on a host of issues, including joint goals and metrics, what data are to be shared, and what to do in case of a disagreement. Processes for providing feedback and reaching consensus are particularly important.

An apparel company based in New Jersey engages in different levels of collaborative planning with several retailers. Goals and agreements between them cover such areas as sales objectives, inventory levels that stores need in order to reach those objectives, and weekly sharing of sales and inventory data. A few of the retailers also share their promotional plans, and some maintain an online database where the apparel company can view relevant data. Key indicators for how the business relationship is performing are also agreed upon, such as a ratio of total sales to average inventory, and profits. There is no set procedure for resolving disputes, but the apparel company says there is an understanding that if demand at a retailer falls short of forecasts, apparel shipments to that retailer will be reduced.

4. Use the Best Technology: Companies should use state-of-the-art technology and standardized data if they're going to get the most out of collaborative forecasting.
There needs to be a central database where different parties can easily store and view the latest sales, inventory and purchasing data. Historical data are important, too, to gauge forecast accuracy over time.

All such information should use language and formats that are easy to understand and use, and products themselves should be tagged with standardized labels, like universal product codes, so there is maximum transparency for everyone involved, including vendors and suppliers.

Demand planners need a system that gathers data from different departments and sources. They also need strong calculating tools that can run a lot of what-if simulations, such as what would happen to sales if the company lowers or raises the price, decreases or increases the advertising budget, introduces new products, enters a new market, or exits the old one.

Supply planners have similar data-gathering and calculation needs. For instance, if a product is expected to be short in supply, should the company put on an extra shift or outsource? Or if supply is expected to exceed demand, should it halt production or build inventory for future use?

Software already exists to help with such tasks. But we expect the emergence of more advanced "what if" simulation software, which will provide faster and more accurate decisions.

5. Focus Where Revenue and Profits Are Greatest: Because resources are limited, companies should focus forecasting on products that yield more revenue and profits.
Trying to track everything is wasteful, especially if the lion's share of profits comes from lines that make up a minority of overall production. So, some companies rank products by A, B and C, with A products yielding the most revenue, or the most profits, and C the least. The companies then focus more of their forecasting resources on those products that produce the most value for the company.

The German-based chemical maker BASF AG adopted an ABC approach in the mid-1990s and now uses it in combination with product-profitability studies to increase forecasting accuracy and improve its product mix, says Alan Milliken, manager of business-process optimization. The method helped one BASF unit improve forecast accuracy by an average of 20% for all of its products, Mr. Milliken says.

6. Link Incentives to Companywide Goals: To ensure better forecast accuracy, incentives and awards for employees should be based on companywide goals, not those of a single department.

If a company rewards production employees on the basis of lean inventory, for example, they could try to lower forecast numbers to maintain as little inventory as possible. Similarly, if salespeople earn bonuses by beating quotas based on sales forecasts, they may supply misleading data in order to keep forecasts low.

The best measures to use as incentives that won't skew anyone's forecasting acumen or pit one department against another are total revenue, without reference to which units contributed what, and profit.

7. Aim for Continuous Improvement: Errors in forecasting can result from bad data, wrong assumptions or a faulty model. So, it's important to conduct a post mortem at the end of each reporting period and take action to correct problems.

Challenge assumptions, processes, technologies and benchmarks. Regularly track all sales and inventory reports. Make a note of changes in the forecasting process as they occur, and of every decision that affects the supply chain. Study the data to see what effects those changes and decisions have.

P&G holds monthly collaborative planning meetings for each of its business units, divided by products, such as oral care or skin care, in every country where they compete. The company conducts annual audits of these processes and gives them scores based on common metrics.

Such strategies mean companies don't have to rely solely on marketing or sales to drive increases in revenue. Supply chains, too, can increase revenue for manufacturers, their suppliers and distributors.

About the authors: Dr. Jain is a professor at St. John's University in New York. Mr. Covas is global innovation diamond manager at Procter & Gamble. They can be reached at [email protected].
http://online.wsj.com