The CPFR model or Collaborative Planning, Forecasting, and Replenishment model is a supply chain collaboration standard. The standard, in part, outlines a collaborative and consensus based approach to forecast generation between supply chain partners.  The forecast is important to ensure as viable of a Sales and Operations plan is put in place as possible to ensure that inventory or capacity is properly positioned to meet customer demands and desired service levels.  Forecasting is an imperfect science. Historic sales information is an important input to forecast generation but there are other key inputs that much be considered. These include:

• New product introductions and/or potential cannibalization between SKUs (Stock Keeping Units or Items)
• New retail sites or closed retail sites
• New customers and/or customer attrition rates
• Future or Historic promotions
• Longer range trends
• Seasonality

Given all of the above, to successfully generate a forecast a Food and Beverage planning organization needs to take a collaborative and cross disciplined approach.  Product hierarchy is important. Depending upon the sku breadth a Food or Beverage company should consider organizing their SKUs by Company, Brand, Product Family, and Products.  Sales hierarchy is important.  Demand should be organized by Sales Channels, Geography, Bill To Customer, Ship To Customer. Finally, dates are important and in some cases such as in the E-Commerce world, time is important.  Time hierarchy can consist of Years, Quarters, Months, Weeks, Days, Time.

To effectively utilize the demand information the sales history needs to be input into a Cube or Pivot Table solution.  To get started Microsoft Excel pivots are fine but as order history accumulates, due to the volume and complexity, it will be necessary to adopt a more formal tool like Outperform Demand Planning.  The formal demand planning tools allow you to look at aggregated demand along any of the dimensions listed above via a Top Down view and also from the detail level up to the aggregation via a Bottoms Up view.  The records consist of both dollars and units.  As such, if aggregation is desired then the units across the SKUs have to be consistent.

Getting sales history correctly aggregated is a big and continuous job.  We at Adroit can help you to get your history reporting cleaned up and consistent, establish data standards, and implement an order history process.

With the order history cleaned up and organized a more sophisticated forecasting and planning tool such as Outperform Demand Planning is used to create a forecast.  The forecast is generated along the exact same hierarchies as the history.  The tool is sophisticated and applies a number of mathematical models to screen out one time exceptions or to use exponential smoothing smooth the forecast. It identifies trends at the SKU and other levels and applies them.  This is known as applying an alpha factor adjustment.  The tool is smart and figures out which forecast technique makes the most sense for the order history profile.  These can include Moving Average, Exponential smoothing, Time Series, Regression, and Multivariate and other models.  We don’t really need to be concerned about it as the tool functions as a black box.

With an initial forecast generated we next need to apply one time future events that we know about. These include new product introductions, new customers, new stores, new websites, promotions, etc. The tool makes it easy to note these and apply them on a Top Down basis.  Typically at this point the forecast is saved as a bench mark.

Next, we need to take apart the forecast to send to various stakeholders with responsibilities either along the product dimensions or customer dimensions for review and approval. It is important that the stakeholders take responsibility for their portion of the demand because the forecast is used to determine purchasing, capacity, production, and inventory levels in the Sales and Operations Planning.

The stakeholders can override the forecast either on a Top Down basis or they can adjust details as well. These are submitted back to the planner for final consolidation and submission to the S&OP team.

It is important that at the same interval the prior forecasts are measured for accuracy and analyzed to determine the reasons for the discrepancy. Understanding what happened is the key to driving improved accuracy.

A solid forecasting process is important to Food and Beverage companies attempting to manage a changing set of products across multiple sales and distribution channels.  We at Adroit can help you to improve your forecasting process and to establish a more robust Sales and Operations Planning process. This in turn will improve your customer service levels while also reducing your on hand and potentially expiring inventory.