Tech2Tech:
Ask the Expert: Taking exception to reports
Is too much information getting you down? Learn the art of exception reporting.

Applied Solutions: Will you survive?
Don't wait until disaster strikes to figure out how to keep your business running.

Insider's Warehouse: The CRM continuum
Successful analytic CRM depends on mature data warehousing capabilities.

Teradata DCM reverses the flow of information in the traditional supply chain, pushing information about demand at the store and SKU level up through the supply chain to vendors and manufacturers.

No inventory reduction effort can be effective without contribution management, a detailed form of analysis that ranks the relative importance of each item at the store level.

REVERSING THE SUPPLY CHAIN

Teradata Demand Chain Management takes the
guesswork out of inventory planning and control.

by Joe McKendrick

LOST REVENUE AND INCREASED COSTS due to overstocks, out-of-stocks, markdowns and out-of-season items are the bane of every retailer's existence. Better organizations manage to stem some of the losses by keeping inventory levels as lean as possible while still being able to fill sudden surges in orders. Yet, no matter how well a retailer manages product flow, writing off hundreds of thousands of dollars' worth of merchandise on an annual basis-and as much on sales due to shortages-is considered a normal part of doing business.

The majority of retailers still conduct inventory forecasting and demand planning using outdated, often home-grown systems that lack forecasting algorithms, exception management and analytical tools and do not consider promotional impact. These systems require a great deal of manual intervention, with users having to obtain data from various siloed systems. Data that's been cobbled together often is full of errors and duplications, both of which limit demand chain efficiency.

In recent years, retailers have turned to technology to automate the process and compensate for imprecise demand forecasts. Basic strategies, which have been implemented with varying degrees of success, include just-in-time inventory, vendor-managed inventory, vendor collaboration schemes and the use of various PC-based forecasting tools. But at the heart of every automation effort is a supply chain management (SCM) system, which supports the ordering and replenishment of goods moving between manufacturers, distributors, warehouses and store outlets.

SCM systems have greatly increased collaboration between retailers and vendors, while reducing human intervention and paperwork. However, SCM as a whole addresses the process from a supply-side perspective and, automated or not, it still puts the onus on retailers to deal with shortages as they arise or dispose of excess merchandise already in the pipeline.

What retailers need is technology that generates forecast data based on information from the front-line of retailing and automatically moves this information upstream to vendors and partners. The Teradata Demand Chain Management (DCM) solution does just that. It helps retailers support collaborative partnerships that enable them and vendors to prepare and react more quickly to changes in consumer demand. Teradata DCM reverses the flow of information in the traditional supply chain, pushing information about demand at the store and SKU (stock keeping unit) level up through the supply chain to vendors and manufacturers.

Buying what you sell
The concept of demand chain management is relatively new. In fact, says David Chapates, Teradata product manager, "the analysts still haven't agreed on a category for demand chain or forecasting and replenishment systems. Demand chain management usually gets lumped into supply chain management. But it's not a supply chain logistical solution. (Instead, it) is based upon demand flow coming from the most detailed level-POS at the store level."

Retail buyers, merchandisers and planners can forecast and manage the flow of goods to their stores based on actual and projected consumer demand for each item. Demand chain management enables you to "buy what you sell, not sell what you buy."

Behind an easy-to-use graphical user interface, Teradata DCM accesses data stored in the Teradata Warehouse. All data generated by Teradata DCM is granular or SKU level, and it can be rolled up into broader merchandise categories or store hierarchies for review and analysis purposes. Teradata DCM modules include Contribution, Demand Forecasting, Seasonal Profile, Promotions Management, Automated Replenishment, Order Forecast Optimizer and Allocation (see "Modular management," page 68).

The Teradata DCM engine contains algorithms that calculate average weekly sales rates to build forecast models. The solution automatically generates results, which retail planners can use to identify the fastest-moving items and reduce out-of-stock situations while trimming inventories of slower-moving items. The software views information within the Teradata Warehouse to automatically calculate next-order series values, by week or by day, for distribution centers and stores.

Scaling the inventory mountain
Teradata DCM is designed primarily for large-scale retailers hoping to trim costs by reducing inventory. The Teradata Warehouse can scale to handle the huge volume of data required for effective analysis and reporting. For example, the process of measuring sales for up to 50,000 SKUs across 1,000 stores requires enormous data storage and processing capacity, and Teradata DCM can accommodate that with the power of the Teradata Warehouse.

The latest release of Teradata DCM is Web-enabled, allowing end users to access forecasting and demand applications and data through a Web-browser front end. This multi-tier architecture enables retailers to quickly deploy the system over a multi-store network via Web server middleware. Web access also enables secure access and collaboration by vendors and trading partners that wish to review demand forecasts.

"If a retailer wants to provide vendors visibility to the information, they can do so with secure user ID and log-in passwords," says Chapates. "Typically, a vendor wants to know 'How much in total are you going to buy from me over a given period?' For a direct-to-store vendor, they're going to want to know how much is required of every product for each store per week." Web access significantly improves that process, saving time and increasing accuracy to boot.

By moving demand forecasting and replenishment down to the store level, large-scale retailers can greatly improve the accuracy of regular and promotional store-SKU forecasts, store-level ordering and promotional planning, and they can confidently align shelf space with sales movement.

Inferior forecasting science and "gut-feel" inventory decision-making have created significant customer service issues across the retail industry. At the store level, as much as 8% of all SKUs and 15% of promoted items are out of stock at any given time, according to industry research. At the same time, overstocks caused by poor forecasting continue to drag down retailer earnings and customer satisfaction rankings.

JCPenney, one of America's largest department store, drugstore, catalog and e-commerce retailers, recently selected Teradata DCM to further improve store-level forecasting and replenishment. The solution offers the retailer tangible results through highly accurate regular and promotional forecasts.

Going against the grain
By building forecasts on factors such as SKU ranking, seasonal profiling, average weekly rate of sale and event management, you can better forecast demand and manage product inventory levels. Some of these factors seem to fly in the face of conventional wisdom, but they can be invaluable with the use of the in-depth analysis tools provided by Teradata DCM.

SKU ranking: No inventory reduction effort can be effective without contribution management, a detailed form of analysis in which the system automatically ranks the relative importance of each item at the store level. By ranking SKUs on a scale of A to E, the system helps differentiate and communicate the overall impact of the products to each location in the retailer's network. Products ranked A and B represent the top 80% of total sales, while Ds and Es collectively only make up 5% of sales.

"The biggest problem is retailers have way too much money tied up and invested in the Ds and Es-the 'dog' items that are not contributing a lot to their business," says Chapates. "When we start working with retailers we typically see over 50% of a retailer's inventory invested in supporting the bottom 5% of their sales." With SKU ranking, a retailer can increase turns by as much as 40% while effectively reducing safety stock and right-sizing inventory levels on the D and E items-all without compromising customer service.

Seasonal profile: Conventional wisdom dictates that seasonality is limited to an isolated segment of merchandise, such as lawnmowers in the summer or snow shovels in the winter. However, most products have significant seasonal selling patterns. When data is scrutinized at the SKU level, seasonal patterns emerge that can be used to build more accurate forecasting models. Identifying the seasonal selling strength of an item (when it sells best, versus how much it sells) delivers more than 50% of the forecast's accuracy at the SKU-location level, a boon to retailers working in geographically and demographically diverse markets.

Average weekly rate of sale: Using de-seasonalized, multiple time-series models to quantify SKU-level demand for a given location lets retailers automatically identify subtle changes in short-term trends. Using this data to select a model for the next forecasting run and adapting the response to the latest trend is essential to responding to customer buying patterns.

Event management: Many retailers do not track promotional uplift and instead use gut feel and intuition to determine the impact of campaigns. Retailers using Teradata DCM's Promotions Management module can accurately predict and account for the increase in sales over and above normal sales. At the same time, they can maximize customer service and merchandise sell-through rates at the individual SKU-location level.

Order forecast: An accurate order forecast significantly improves the synchronization of the distribution center/warehouse with the requirements of individual stores, providing the ability for retailers to collaborate with vendors and, ultimately, shift from a supply chain to a demand chain.

Asking "what if"
Even with hard data, a forecast is still just that-an estimate of some future event. Teradata DCM's Promotion Management module includes a feature that enables retail planners to run what-if simulations based on the parameters attributed to an event. The event planner feeds the simulation with promotion goals, such as sales and gross margin, and with factors such as media type. Other information related to the event, such as pricing or discounts, locations, dates and displays, is also included in the simulation.

Teradata DCM runs an Automatic Event Uplift Engine against these variables-as well as the retailer's sales and event history-to calculate the promotional uplift coefficient. The planner can manipulate the various parameters until suitable targets are reached. The end-user can then "lock in" the simulation as an executable forecast, and the additional uplift requirements are automatically reflected in pending forecast and replenishment orders.

The Teradata Demand Chain Management solution represents a dramatic new approach to building more accurate demand forecasts, as well as enabling more collaborative relationships with vendors and suppliers. It focuses on consumer demand and creates a new flow of store- and SKU-level information back through the supply chain, enabling retailers to truly buy only what they will sell. T

Joe McKendrick, research consultant and author, contributes to Evans Data Corp., IDC, Gartner and Faulkner Information Systems, as well as journals such as Database Trends & Applications and Enterprise Systems.

ILLUSTRATIONS BY JOYCE HESSELBERTH