Master Data Model: Creating a Single Language for Retailers

Centralized and uniform master data is the foundation for creating an integrated, flexible and adaptable supply chain

A famous African parable tells of three blind men, from different tribes and speaking different languages, who were each asked to describe an elephant. However, each man had only experienced a different, singular part of the animal. One of the men, after touching the elephant's leg, concluded that the elephant is like a large tree without any branches. The second man, after feeling the trunk of the elephant declared, "No, it's nothing like a tree it is like a mighty snake." Finally, the third man, who had experienced only the tusk of the elephant, observed that the elephant is nothing like a tree or a snake it is long and smooth and sharp. "Clearly," said the man, "an elephant is like a spear."

Of course each one of the men accurately described their understanding of what an elephant is based on their experience, but finally none of them was able to accurately describe an elephant.

Piecing Together the Supply Chain Vision

Many retail organizations today operate in a similar manner. Each business unit has a view of their part of the organization, but it is often difficult to form a common or a complete understanding of the business because there is no way to link the disparate views. Commonly, the capability to form a coherent understanding is lost because the master data (the foundational data that is common to every department in the organization like sites, vendors and items) is incomplete, inaccurate or not aligned between systems.

One retail manager described it this way: "It's like you're in a room with peep holes all the way around and you need a panoramic view to understand what's happening in the business. So you begin looking through the peep holes, one at a time, to see what's happening out there, and finally you are able to piece together what you think is most of the picture. However, you realize that it's not today's picture but that it's yesterday's picture, and then you start the process all over again."

It has often been said that good decision making requires good information delivered to the right people at the right time. Taking this thought one level deeper, it can also be said that the bedrock of good information is well organized and coherent master data (also called referential data). These days there is very little room for error in a highly competitive retail industry with paper-thin margins and increasingly higher costs of doing business. Companies looking for every small advantage must look at the foundation of their decision making (master data) and ensure that they are not entering the competition with an avoidable and all too correctable handicap.

Creating Understanding through Common Language

One can imagine that some other person, interviewing each man in the African parable, could begin to cobble together a pretty accurate picture of what really is an elephant. But what might it be like if the men used different words to describe the same thing? A leg might be called an ear, a tusk might be called a tail and an eye might be called a trunk. The picture that would emerge in such a case would be strange indeed. Surprisingly, plenty of retailers do something very similar to this when trying to form a global picture of their supply chain.

To illustrate this confusion of language, let's look at a situation that will be familiar to many. Logic would dictate that a can of corn in Kansas is the same as a can of corn in Florida. At many retailers, however, the merchandise system in Kansas does not communicate with the merchandise system in Florida. Identical items are added to each system using the next available warehouse item number, and these numbers are usually not the same. For a retailer to determine how much corn is being moved in the enterprise can be a time-consuming endeavor. By using the case UPC they can begin to piece together a picture of movements into and out of the distribution center (DC). However, to see how that relates to unit sales from the point of sale requires a lot more detective work. With this model there is no good way to directly translate consumer demand to what is being ordered into the warehouse. For these organizations forecast optimization is difficult at best and impossible at worst.

Every department and corresponding information system that supports the retail operation requires master data. Merchandising, category management, sales, warehouse operations, transportation and accounting have different but complementary roles to play. In many cases master data is created in each system to support the function of that particular department without serious consideration of how that information translates into other systems or other departments.

Creating a truly integrated supply chain with visibility at every touch point can occur when all of the master data components are centrally created and maintained in a single system and fed from there to other business areas that use the data. It will be understood the same way not only in Kansas and Florida but also in operations, merchandising, sales and accounting.

Creating Data with Integrity

A centralized master data model can deliver sustained benefits, including improving data integrity. When the same data is entered by different people into different systems, over time it is likely that the data becomes disjointed and unreliable. Once the data loses credibility in an organization it can be difficult to restore. What often happens is that each group or department has their own criteria for success, and measuring performance can lead to contentious debates and ambiguous results. However, when everyone is using the same data to describe the same thing it removes much of the subjective measures of success, makes it easier to manage expectations and to create accountability within an organization.

In addition, data integrity issues can lead to lost business opportunities. For example, let's say a buyer in Texas is moving a large volume of product from the vendor "Jeff and Joe's Country Cookin'" and is able to negotiate a favorable deal. The buyer from the same company in Arizona is doing less volume of the same product, but in Arizona the item number and the vendor number are different, and the vendor description is "J&J Country Cooking." Because the data is not uniform between the systems, the category manager may lose negotiating power and is likely leaving money on the table.

Creating a Flexible Supply Chain

A centralized master data model is also the staring point for creating a flexible and adaptable supply chain. During operational interruptions, for many companies, it can be the difference between survival and ruin.

Take, for example, a retailer with a distribution center in New Orleans that supplies Baton Rouge and the southern Gulf Coast. Suddenly, a hurricane strikes that shuts down operations for an extended period of time. The retailer has a couple of options: shift distribution for those stores to another facility, or leave the store shelves empty until the DC is operational again. Of course any retailer would choose option one, but all too many of them have a master data model that would severely limit their ability to respond. The store's order code, for many retailers, is the combination of item code and DC code, and changing the distribution channel requires changing the order guides, which for most organizations is a Herculean task.

Changing the master data model to facilitate abrupt changes in sourcing and distribution will require some changes in the way retailers think about their data. The key is to uncouple the raw data from anything else. An item is an item by itself, a supplier is a supplier by itself and a site is a site by itself. In order to be useful, of course, these things must be linked together, but by storing them as completely separate and unrelated pieces of data one can easily and quickly mix and match them to any configuration that is required for the operational situation.

Many merchandising systems make the mistake of forcing operational decisions to be made upfront and inextricably linking the master data to those decisions, which severely limits the retailer's options. The optimum model is to limit the information that is stored at the root level to data that is required to accurately describe the particular thing.

The item root, for example, should not have to account for sourcing, costing, pricing or distribution, because all of these things can change over time. The item itself consists of type (standard, fresh, recipe), dimension, configuration and a couple of other specific attributes and should stand alone as a completely independent piece of data. Once the items, suppliers, contracts and sites have been uncoupled then it is possible to begin linking the data to support the operations. It is also possible to reconfigure the links to support changing conditions. No longer will one think about DSD items and warehouse items, but rather one will think about an item the can be sourced from anywhere to anywhere at the touch of a button.

The retail supply chain, like the elephant, has many separate but interrelated parts, and none of these parts, taken out of context, can complete the picture. In order to enable a complete view of the retail supply chain, the master data must be accurate and consistent between systems. Ultimately, centralized and uniform master data is the foundation for creating an integrated, flexible and adaptable supply chain, making it possible to maintain perpetual inventory, optimize ordering, run more effective promotions and create a more efficient, more profitable organization.

About the Authors: Neil Thall is CEO and Derek Corrick is Senior Implementation Consultant at Aldata Solution, a global provider of retail software solutions.