Skip navigation

Cleaning Up

On AMR Research's list of the applications deemed most important in 2008 by fast-moving-consumer-goods retailers (operators of supermarkets, drug stores and convenience stores), the application that got the highest percentage of votes (19%) was master data management. In 2007, it was near the bottom of the list. Master data management helps retailers and manufacturers maintain current, clean and accurate

On AMR Research's list of the applications deemed most important in 2008 by fast-moving-consumer-goods retailers (operators of supermarkets, drug stores and convenience stores), the application that got the highest percentage of votes (19%) was master data management. In 2007, it was near the bottom of the list.

Master data management helps retailers and manufacturers maintain current, clean and accurate product data that can be aligned consistently across all parts of an organization. But why the sudden interest in this technology?

“In order to use more sophisticated technology like demand forecasting, pricing, space management and perpetual inventory, you need a solid foundation of [product] data,” said Michael Griswold, research director, retail, for AMR Research, Boston. “What has risen to the top is the need to get the underlying data synched, organized and maintained.”

Synchronization of item data between retailers and manufacturers via the Global Data Synchronization Network (GDSN) has helped many retailers begin the process of getting their product data in order. Retailers know that data synchronized through the GDSN is formatted according to prescribed standards — for product name, description, dimensions and other attributes — that all manufacturers and retailers in the network follow.

But as retailers that have pioneered data synchronization — such as Wegmans Food Markets and Supervalu — have discovered, it's not enough for product data to be synchronized; it also has to be accurate. That is, a product's dimensions, as well as logistical information such as pack size and case dimensions, have to reflect physical reality. If they don't, a retailer could end up squandering millions of dollars in transportation and warehouse inefficiencies, to say nothing of billing errors and other logjams.

“Synchronizing bad data doesn't do us any good,” said Connie Vandervort, e-business director for the Army & Air Force Exchange Service (AAFES), Dallas.

The other issue is that even when data is synchronized between a retailer and its manufacturers, it is not always synchronized among divisions inside the retailer, such as distribution, store operations, merchandising, marketing and the website. “Data synchronization facilitates consistency between suppliers and retailers, but that's just 20% of the problem,” said Griswold. “The other 80% lives in the retailer.”

One of the tools being used to address data accuracy and consistency is the product information management (PIM) system, which is seen as a form of master data management as well as a tool that prepares data to enter the master data management system or other data repository.

Retailers such as Ahold USA and Price Chopper are implementing PIM solutions, which tend to be highly customized. Ahold USA, Quincy, Mass., is developing a new IT architecture based on applications from Oracle as well as a PIM from GXS. Price Chopper, Schenectady, N.Y., plans to leverage a PIM from Aldata as part of the implementation of the vendor's GOLD retail supply chain suite. These retailers reflect the tendency to adopt PIM along with other supply chain technology. Other PIM suppliers include IBM, i2 and Soft Solutions.


AAFES, which operates more than 3,000 retail outlets ranging from department and specialty stores to convenience stores across the U.S. and around the world, started deploying a PIM from GXS, Gaithersburg, Md., in 2005.

AAFES uses the PIM to “cleanse” product data that it imports through the GDSN, said Vandervort. In particular, the system helps vet the accuracy of logistical data such as case dimensions, pack size, case cube and pallet tie-high. AAFES also conducts random physical audits of container characteristics and asks suppliers to verify the accuracy of their data before downloading it.

“Dimensional data is the big one,” she said. “I've heard horror stories from other retailers talking about how their dimensional data is all wrong.” Once the dimensional data is cleansed, she added, “we cube out space in trucks better, which saves on transportation, and we save space in our warehouses.”

AAFES has finished cleansing product data from 556 suppliers out of a total of more than 5,000, so there are “thousands more to go,” said Vandervort. Those 556 suppliers all supply data via the GDSN, which is AAFES' preferred method of receiving product data, but it has established a portal for small suppliers and others that don't use the GDSN.

Data cleansing has not happened as quickly as Vandervort had hoped, but that is not the fault of the technology, she said. Other factors, such as the merger in 2005 of UCCnet and Transora to form the 1SYNC data pool and the change in business practices at AAFES, have also affected the speed of implementation.

A supplier synchronizing data via the GDSN with AAFES publishes data for all of the items it supplies to AAFES. The data is staged in the PIM, which overlays dimensional data in AAFES' system on the supplier's data. Any inconsistency found by the PIM goes on a report that is brought by AAFES' “cleaners” — people whom Griswold called “data stewards” — to an AAFES buyer and the supplier “to try to figure out what's right,” said Vandervort. AAFES also attaches to product data its own user-defined attributes, which will eventually be made part of the PIM.

Though the process can be slow, “it's worth doing, because data needs to be cleansed,” she said. “We find inaccuracies on our side and on the suppliers' side.” Once the data has been corrected, it is channeled to AAFES' Oracle Retail merchandising system, which feeds it to other databases in AAFES, ensuring that the data is the same across divisions. Subsequently, any new items or changes to items are automatically staged and processed via the PIM before moving to Oracle, increasing speed-to-market.

The PIM also enables AAFES to dispense with keying in the item information coming from suppliers — a long and mistake-prone process, especially with more than 1.8 million stockkeeping units in its system. “The manual process would be unreal without the PIM,” Vandervort said. As it stands, AAFES employs 40 people who key in new items and changes. “Our intent is to bring most of that through the GDSN so we won't have that expense,” she said. At that point, some keypunchers will be reassigned as data cleansers.

Incorrectly keyed information used in purchase orders can play havoc across the supply chain. “The errors follow you through the system — in-advance ship notices, finance and accounting,” she said. “It can be very expensive.”

Vandervort could not say what the investment in the PIM has been. AAFES has reduced its costs by having GXS host its PIM, though at some point the retailer will bring the system inside its IT infrastructure. “With our Oracle Retail implementation, we tied up IT resources, so the quickest way to get PIM functionality was for [GXS] to host it,” she said.

The cost of an internal PIM for a “good-size company” can run between “a couple of million dollars and $10 million or more,” said Bryan Larkin, GXS' director of strategy and marketing for retail and consumer products.

AAFES' next step will be to engage in price (product cost) synchronization, which last year was standardized by GS1, Brussels, the parent organization to the GDSN. AAFES was one of a handful of retailers who participated in pilots supporting the development of the price standard.