After years of effort, the retail and consumer packaged goods industries are close to correcting one of their most costly and intransigent supply-chain problems: mismatched and out-of-date item data.

For a wide variety of reasons, from sloppy keyboard entries to major corporate mergers, trading partners have long endured the inefficiencies that come when a supplier's item number and description do not match that used by a buyer.

"Over time people have gotten used to this huge business of managing trade settlements," says Andrew White, research analyst with Gartner Group, Stamford, Conn. "It is now part of our daily lives and we have software and budgets to cope with this rubbish. And why do we have all these problems? The root cause is that one enterprise doesn't have the same product data as another enterprise."

According to a study by A.T. Kearney for the grocery industry, which is widely accepted by a broad range of retailers and manufacturers, "dirty data" issues are at the bottom of a host of dismal operating statistics:

• $40bn or 3.5 percent of sales lost each year due to supply-chain information inefficiencies that lead to stock-outs.

• A 30 percent error rate in item data catalogs used by retailers and manufacturers for stock replenishment. Each error costs $60 to $80 to address.

• An average of 25 minutes per SKU each year spent to manually correct out-of-synch information.

• Errors in 60 percent of all invoices generated, with each error costing between $40 and $400 to reconcile.

• Deductions taken on 43 percent of all invoices because of inconsistencies.

• An average of four weeks to roll out a new product, in large part due to the inefficient and error-prone approaches for the exchange and updating of new item information in buyer and seller systems.

Getting both sides of a transaction to use the same item data - a process known as global data synchronization - can reduce these drains on efficiency and profits to the tune of about $1m for every $1b in sales, according to A.T. Kearney. In a subsequent study based on several proof-of-concept pilot projects, Kearney enumerated specific benefits for retailers and manufacturers. (See box).

In addition to these direct dollar savings, experts say, data synchronization is foundational to many other initiatives and systems. "Benefits of Vendor-Managed Inventory and Collaborative Planning, Forecasting and Replenishment depend on accurate product information," says Eric Austvold, research director for enterprise applications and technology strategies at AMR Research, Boston. "Data synchronization is the basic enabler for making collaboration really work."

This is demonstrated by a major CPG company that had made a huge investment of time and money to enable it to receive 75 percent of its inbound orders electronically, says John Stelzer, director of industry development at Sterling Commerce, Columbus, Ohio. "That is a very impressive statistic," says Stelzer, "but half of these orders were getting kicked out with exceptions that required manual intervention. So all of this company's investment up to that point was being undermined by bad data."

Data synchronization also is a necessary first step for next-generation collaboration technologies like scan-based trading, the electronic product code (ePC) and radio frequency identification (RFID).

"We feel that as companies look to more effectively execute on brand strategy and compete in the global marketplace, data synchronization is a fundamental requirement," says Rene d'Ouville, vice president of product strategy at QRS, Richmond, Calif. "It has a return-on-investment in and of itself."

Jeff Oddo, manager of corporate communications at UCCnet, Lawrenceville, N.J., agrees. "Getting data that is good, solid, clean and based on standards is absolutely foundational to other processes," he says. "If you have something at the foundational level that is wrong, there is a domino effect that impacts every other step that builds on top of it, all the way down the supply chain."

Solution Evolution
So why hasn't a problem with so obvious a solution been fixed before now? Two key reasons: the lack of global standards and the lack of an easy and inexpensive means to connect and transmit data. The internet and XML (extensible markup language) solved the latter problem; the first has simply taken time.

In 1998 six companies - Frito Lay, Kroger, Procter & Gamble, Ralston Purina, SUPERVALU, and Wegmans Food Markets - formed an ad hoc committee to further explore the concept of an internet-based, electronic trading community. Coinciding with this event, the board of governors of the Uniform Code Council charged that industry standards organization to become more actively involved in electronic commerce solutions. The two initiatives soon merged into one to create a not-for-profit subsidiary of the UCC, called UCCnet.

	"Just to find all the instances of item information is like a giant scavenger hunt." - John Stelzer of Sterling Commerce

Under the auspices of UCCnet, the six founding companies completed several proof-of-concept pilots to test new business procedures and processes for data synchronization. In July 2000, UCCnet formally launched its global data synchronization services. At first few companies signed up, partly because some standards and other issues still were in flux, but after several industry groups called for action in 2002, the pace picked up. As of the first quarter of this year, more than 3,000 companies have joined, representing a wide range of sectors - though only a small percentage of these are actually exchanging data at this point. Grocery companies and their suppliers continue to lead the way in numbers. Following close behind are retailers such as Lowes, Home Depot and Ace Hardware and general merchandisers, most notably Wal-Mart. Most of these leading retailers have mandated that their suppliers be ready to synchronize data by 2005.

Greg Lenard, director of inventory control at Ace Hardware, has worked closely with the group establishing product descriptions for the hard goods sector. "The Kearney study nailed it right on the head in terms of the inefficiencies," says Lenard. "A lot of people don't believe the numbers because they are so large - and it's true that we may not see full results until the number of companies participating gets to critical mass - but I think we already are at critical mass in terms of people getting ready. Now it is just a matter of implementation."

How It Works
So how does data synchronization actually work? The basic premise is very simple, says Oddo. "I want my description of an item and your description of an item to look exactly the same, in every instance, all the time." When that happens, most errors, disputes and deductions will disappear.

To get there, buyers and sellers first had to agree on how to describe items. "The biggest issue was that there was no common or standard way to describe the attributes of a product," says AMR's Austvold. "Things like color and size are pretty basic, but you also have to get into lots of other descriptors, like what kind of package the item is in and the dimensions and weight of each package. How many are in a case and when you aggregate cases, how many make a pallet?" Various standards committees, in conjunction with UCCnet, have been busy at work putting these in place and establishing standard XML schemas for sending and receiving the information. To date, 153 attributes have been identified that can be used for item descriptions. Separate standards are being developed for synchronizing prices and promotions, and these will be rolled out over the next several months.

Once trading partners agree on attributes, they then can use a global key to identify products, Oddo explains. In this case the key is the 14-digit Global Trade Item Number or GTIN.

The GTIN is the identifier for trade items that is issued by EAN.UCC (the joint venture between UCC and its European counterpart, EAN). The 14-digit data structure allows this identifier to accommodate all barcodes now issued by these bodies - the 12-digit U.S. version and the 13- and 8-digit versions used in Europe. This is important because data synchronization clearly is intended to be a global effort. One digit of the GTIN is a check digit used to ensure data integrity, and the others are variously divided to represent a manufacturer prefix and an item reference number.

The rules for assigning GTINs ensure that every variation of an item (such as a single Coke or a six-pack) is allocated a single reference number that is globally unique. The GTIN numbering structure does not contain any meaningful information in itself. GTINs are simple pointers to database information that can be directly used in any company and in any country.

And that is where the global item registry and data pools come in. UCCnet hosts the global registry, which will formally launch in July. Every registered item and its GTIN will be kept here along with a number identifying the seller (the Global Location Number) and a list of country codes in which the product is available from this source. The more complete product information, with attributes, will be stored in a data pool. Currently, UCCnet hosts a data pool and will continue to do so separate from the registry, but others also are in this business. Transora and WorldWide Retail Exchange are both providing data pool services to retailers and manufacturers. QRS Catalog and Global eXchange Services, Gaithersburg, Md., are other examples of data pools in the U.S., and there are many other data pools, often country-specific, throughout the world.

Publish & Subscribe
Data pools can specialize in source companies that send data, in receiver companies that accept data, or they can provide services to both. Source companies publish their data to the pool and recipient companies subscribe to all or some of that data. Data pools are, in effect, inter-operable hubs that validate data to ensure it is standards compliant. And they enable the transfer of validated data between publishers and subscribers.

Companies cannot actually interact with the global registry themselves, but must register their products through a data pool that is part of the Global Data Synchronization Network.

Nick Parnaby, global marketing director for the WorldWide Retail Exchange, Alexandria, Va., likens a data pool to AOL or Mindspring on the internet because "it provides a single point of access, allowing you to pick your way into the network." For example, he says, a retailer might enter the network through WorldWide Retail Exchange to synchronize data on specific products from Procter & Gamble. WWRE contacts the registry to look up the item information, he says. It finds that the information is available through the Transora data pool. WWRE and Transora are connected, so WWRE can point to the requested data and flow it back to the recipient.

To be UCCnet certified, data pools must meet inter-operability standards. This means that all data pools in the Global Data Synchronization Network are required to show they can exchange information with other data pools.

Over time data pools will become very commoditized, says Parnaby. He again uses the AOL analogy. "Compare AOL to a service provider that just gives you access to the web," he says. "AOL competes on all content and community features - it can't compete on just getting on the internet. This will be the same for data pools in the future, but right now the race is on to see who will sign up the most companies."

Managing Change
Data pools also provide important messaging services between source and recipient firms that confirm that data has been successfully received. This latter is key to making sure that data, once synchronized, stays that way.

"The data synchronization process brings not only agreement on one version of the truth, but a continuous refreshing of that information whenever anything changes," says Oddo. "This is very important from a supply-chain perspective since things are always changing."

Ace Hardware's Lenard notes that one of these changes is discontinued items, which often are not communicated to retailers in a timely way. "This is one of the wins that is coming out loud and clear for us," he says. "In the past, here we would be expecting to replenish an item that the manufacturer isn't even making anymore. That is pretty important information for us to know and the sooner we know it, the sooner we can stop disappointing customers."

When data synchronization is in place, any product change will automatically be sent to the data pool, which will publish it to subscribers.

Similarly, says Kevin Babb, product manager at software provider Inovis, Atlanta, data synchronization will make it much easier to add new products to retailers' master item lists. "When companies are introducing a new product, they want to get it to market as quickly as possible," he says. "Now there often are delays around notifying retailers as to what this item is, its attributes, and other information they need to start ordering it, and those delays cost a lot of money."

Most of the hard work of data synchronization is going on now behind company firewalls, getting data and systems ready to synchronize. This takes most companies longer than they expect, says Austvold, so any company wanting to meet a January 2005 compliance date already should have this effort well under way. Essentially, internal synchronization involves finding all the item data housed in different systems within the four walls, aggregating it and cleansing it so there is a single, clear and accurate description for each item.

"Just to find all the instances of item information within the four walls is like a massive scavenger hunt at most companies," says Stelzer. Cleansing the data is largely a matter of "good old-fashioned elbow grease," he says. "Even if you have technology to help you find exceptions, you need humans to make a decision."

Software providers are working on tools in this area, however. Inovis has a product called IPNet that employs a hybrid process to review and cleanse data, says Babb. "Experience thus far has shown that there are typically two errors that occur in most of the data we have reviewed - either missing data or malformed data." Malformed data can usually be corrected without any outside assistance, he says. "Missing data is more time-consuming to correct and involves requests to the manufacturer for assistance."

Once the data is found and cleansed, procedures need to be put in place to make sure that future changes are controlled and communicated to every system that uses the information.

"Companies have the option of putting the data all in one place and forcing all systems to use that database," says Stelzer, "but that is fairly cathartic. The reason the data is in three places now probably is because the applications need it there. So the idea is to control and coordinate changes so that when a change to the item list is made, other systems are automatically updated and you stay in sync with yourself and your partners."

Once cleaned and validated, data is sent to a synchronization engine, either in-house or at a data pool, to be put into the proper format for transmission.

Retailers also must go through a preparation process behind their firewalls. They too they must find all instances of product data in their systems and make sure that synchronized data, and all subsequent changes, reaches every system.

Help from Software
Sterling's Gentran Integration Suite serves both purposes, says Stelzer. "Gentran has the ability to link multiple disparate systems and to put in various process rules and execution steps that can monitor a particular data store, detect when change has happened there and update all other systems with that information."

Other software companies also offer solutions in this area. Velosel, Santa Clara, Calif., helps manufacturers extract item data from a range of sources, cleanse and enhance it with the additional details required for synchronization and then load and synchronize with trading partners. "The whole idea of data synchronization for manufacturers is in taking the view that we are not just sending 150-plus attributes across the wire, but we are trying to figure out how to make it useful for the entire enterprise" says Russ Henry, senior vice president of marketing. "Where we have had success is in taking bits and pieces of our customer's business - stratifying it either by product line or geography - and improving the processes, making that data more complete and accurate. Customers that approach it in that way have more early success and build momentum for keeping on."

Global eXchange Services last year enhanced its offering with the acquisition of Hot Commerce, says Alex Schumacher, retail vertical marketing manager. "We can manage internal synchronization as well as providing a full range of solutions for external synchronization to manage the business-to-business data flow," he says. GSX also has a product focused at small to mid-sized companies that want to synchronize product data with trading partners. "This allows them to rapidly do the catalog aggregation and simple syndication, in addition to leveraging more standardized compliance or validation rules," he says.

Inovis, whose roots are in the world of EDI translation and messaging, has introduced BizCatalog and BizSync to enable data synchronization. These solutions are designed to help companies of any size that need a rapid implementation to meet retailers' mandates, says Babb.

QRS recently added a product information management solution called Impact to its existing catalog solution. Impact is built around the concept of a centralized data repository and is designed specifically for brand marketers, suppliers and retailers, says d'Ouville. "We think a centralized system is the best way to streamline information because you don't have to replicate it or store it in other systems," she says.

Most experts predict that data synchronization will be common operating practice for most retailers and manufacturers in the U.S. by the end of next year. Global participation is still lagging, however, and that will be the next challenge.

Austvold emphasizes, however, that companies can realize substantial benefits now from moving ahead. "Our research shows that manufacturers that take control of content and invest in data synchronization garner immediate cost savings and competitive advantage," he says.

Summary of Benefits

Manufacturers • 3% to 5% reduction in shelf out-of-stocks. • Two-week reduction in speed-to-market for new items (i.e., 14 extra days' sales of faster-moving items). • 7% to 13% reduction in sales force time spent communicating basic item information to customers, following up, resolving queries, etc. • Reduction in call center and web site queries regarding basic item information. • 5% to 10% reduction in sales force and accounting time spent dealing with invoice disputes. • Reduction in invoice write-offs incurred as a result of data discrepancies. • Elimination of basic item data errors, currently found in up to 8% of total purchase orders. • 0.2% to 0.7% reduction in outbound logistics costs. • 0.5% reduction in inventory. Retailers •3% to 5% reduction in shelf out-of-stocks. • Two-week reduction in speed-to-market for new items (i.e, 14 extra days' sales of faster-moving items). • 10,000 to 30,000 hours saved in store labor costs resulting from shelf-tag and scan errors. • 5,000 to 10,000 hours saved in merchandizing and data entry time dealing with new item introductions and updates. • 1,000 to 2,000 hours saved in finance time dealing with invoice disputes related to basic item information. • Reduction in invoice auditor fees. • 0.5% to 1% reduction in inbound freight costs. • 1,000 to 2,000 hours saved in dealing with item discrepancies. • 1% reduction in inventory. Source: A.T. Kearney, "Data Synchronization Proof of Concept: Case Studies from Leading Manufacturers and Retailers"