You are here
Call it what you will -- scan-based trading, price optimization, data mining, CPFR, CRM, loyalty marketing, supply chain management systems, RFID, work force management software -- anything technology-related in retail has two main goals: collecting data and turning that data into actionable information. Regardless of what flows from this information -- whether it's increased customer service, a streamlined supply chain, or reduction in shrink or out-of-stocks -- it translates to one thing: making money.
The center of all this is the transaction log, or T-log, which captures all of the data from the point of sale. "No matter how sophisticated we get, the heart of all advanced analysis is the T-log," says Rob Berman, v.p., retail for Dayton, Ohio-based Teradata, which provides enterprise data warehousing and enterprise analytic technologies and services. "I would contend that if a retailer has a scanner and a point-of-sale system, they have enough to start with. It is there they capture products sold, store location, time, customer information, cashier information, how they buy, how they pay; it is good, clean data."
Indeed, it's this information that most marketing, mining, and loyalty programs begin with in their analysis. "From this information you learn a great deal about your customers and their shopping habits," Berman says. "You learn what segment they fall under, if they are price shoppers or cherry pickers, loyalty customers or high-volume customers; if you develop 10 or 15 segments, you have most of the customers you want falling within them.
"Just take a look at your average transaction receipt. You have the product, store location, time of the transaction, customer information, how they buy, how they pay, who checked them out -- there is such a wealth of information that you start out with from the point of sale alone."
How they do it
Catalina Marketing of St. Petersburg, Fla. is another example of a company that helps retailers make the most of the information captured at the POS. With a network of more than 21,000 grocery stores streaming in approximately 250 million transactions a week, the company has developed one of the six largest databases in the world. It contains the purchase histories of more than 100 million household IDs. Based on this information, the company offers grocers a variety of marketing programs.
Using T-log data, there are three major groupings of solutions grocers can benefit by addressing, according to Eric Williams, Catalina's c.i.o. One is category management—taking the data aggregate by product groupings of the same type. The second grouping is customer lifestyle, such as families, single men, or single women. The third group consists of club members, such as members of a wine club or cheese club.
Under the category management groupings, grocers can make sure that all areas of the store are being shopped, and can make adjustments in marketing plans to boost sales in areas where the data indicates weakness. "You want to measure your top category sales against what your top shoppers are doing, since the best shoppers usually lead the trends," Williams says. "Usually, if a category is down, the stores' best shoppers are not shopping there, so you want to develop specific promotions to get them into those areas. Once their sales rise, so will the category's."
While applications for loyalty programs often capture lifestyle data, grocers don't have to depend on them to determine this grouping of customers, Williams notes. "They can use a snapshot of POS data, paying close attention to specific items that signal whether a customer is a member of a particular lifestyle group," he says. "For example, Nestle Hot Pockets are usually bought by people with young children, or families that are always on the go. By flagging this product as well as some other identifying products, the grocer can also get a good idea of who is shopping in their stores."
Club programs, such as the wine clubs offered by many grocery chains, are another way of capturing detailed customer data, Williams says, because members often opt in and are very active participants. "Whenever you have a club of some sort, you capture extremely detailed and valuable information about a segment of your shoppers," he says.
But even the T-log can have bad data creep in, especially when the grocer sells a variety of items without UPC codes, such as produce. "This highlights another area that can be tracked by T-log information, and that is labor," Teradata's Berman says. "One cashier, in an effort to appear to have faster throughput, may actually cost the company sales by skipping important steps during the transaction."
Berman uses apples as an example. "All apples look the same to me," he says. "A Rome apple is similar to a Red Delicious, but when you get to organic vs. nonorganic, even a produce manager may not be able to tell the difference without some sort of labeling. In an effort to get customers through the checkout faster, a cashier may put every apple through as a Red Delicious. But the other apples cost twice as much, in some cases. Multiply this by a few hundred stores and it turns into a great deal of money the chain is not getting. While that cashier may on the surface look as if he is more productive than others, he may actually be less productive on a per-transaction basis. All of that information can be determined by the T-log."
According to Ernst & Young's 2002 Loss Prevention Survey, POS exception reporting -- in which certain events are red-flagged, such as entering only one type of apple repeatedly -- was among the most effective loss-prevention procedures used by retailers. Loss-prevention technology companies such as Trax Software & Consulting in Scottsdale, Ariz., develop software that helps spot key performance indicators in POS data that point toward potential shrink, such as value per item, customers per no sale, negative transaction activity, items per minute, and sales per hour.
Even if the data coming from the T-log is perfect, it's only part of the equation. "Where things start getting tough is when you start incorporating data from other sources outside the T-log," Teradata's Berman says. "It may be coming from a mainframe, it might be coming from a data mart or from a vendor's system."
In addition, with all the mergers and acquisitions in the industry, it's not unusual for grocers to have numerous data warehouses with varying data representing similar products and categories.
Indeed, ranking No. 1 on Forrester Research, Inc.'s Ten Retail Technology Predictions for 2004 is this forecast: Internal use of data will dominate the IT agenda.
According to the report, released in January, "Most retailers don't suffer from a lack of data, but many suffer from a lack of clean, well-organized data and/or the skills or tools to take advantage of whatever good data they have. The year 2004 will be the year that retailers zero in on what matters most, and IT and business leaders will work side by side to improve how they capture, analyze, and make decisions from increasingly accurate and centralized data."
Inaccurate data makes it extremely difficult for grocers to determine the most important -- and perhaps the most elusive -- of all figures: net profitability. "It takes a lot of time and energy to determine profitability," says Pam Stegeman, v.p. of supply chain and technology for the Grocery Manufacturers of America. "You have the sales from the point-of-sale data, but the costs are a completely different issue, and when you have the razor-thin margins grocers do, any reduction of cost impacts profitability."
Because of its effect across the entire organization, bad data can have a tremendous impact on cost. "For all the money a company can invest in integration and automation, all that is needed is a small amount of bad data to undermine the entire system," says John Stelzer, director of industry development for Sterling Commerce, a Dublin, Ohio-based provider of business-to-business e-commerce solutions. "Errors processed are still errors. It's like being dumb faster."
According to A.T. Kearney:
-$40 billion, or 3.5 percent, of total retail sales are lost each year due to supply chain information inefficiencies.
-30 percent of item data in retail catalogs is in error, and each error costs trading partners $60 to $80.
-Trading partners perform 25 minutes of manual cleansing per SKU per year.
-60 percent of all invoices generated have errors, 43 percent of invoices result in deductions, and each invoice error costs $40 to $400 to reconcile.
Just a simple error on one purchase order can send a wave of data confusion throughout a grocery organization -- from shipping and receiving to buying, merchandising, and back-room inventory -- and have detrimental impacts on customer service and sales.
Just what causes this bad data? "It can come from a variety of sources," Stelzer says. "Often it is a keying error, especially considering the way the numbers usually first get into the systems of the retailer. It can be faxed, e-mailed, dropped off in person by the supplier's rep as he is heading to the main office. That item then might get passed to several people in the retail organization who key-enter that information into various applications in the retailer's system."
Once it's in the system, however, other things can happen to the data. "There may be a new feature added, and perhaps the retailer changes how they track that SKU," says Stelzer. "For example, they may be tracking a set of batteries as part of their flashlight display instead of a battery display, so they kind of look at it as a different retailer SKU, because it is how they are merchandising it. They may have assigned a different number to it, and as they assigned that number to it for internal tracking of that retailer SKU, they may not appropriately map that back to the appropriate UPC number of the product itself."
Once the data is in the system, the computer just assumes it's true.
Access to data may be another issue in allowing bad data into the retail organization. For example, a grocer selling television sets or other boxed items may have an item in a database with the boxed dimensions entered into the system to be used by shipping and inventory for storage purposes. Someone from merchandising may be setting up a planogram and may change the dimensions to those of the product outside the box -- for merchandising purposes. If logistics goes back into the system and uses the changed data for transportation purposes, the amount of space needed will be underestimated. That could result in unexpected costs for an additional less-than-truckload delivery.
"One retail executive we have worked with was able to track how many people have access to change one particular field in his system, when the field was changed, and what values were changed to," Sterling's Stelzer says. "He found that three different people changed the data attribute for one item -- unbeknownst to each other -- within a 24-hour period."
To determine what these kinds of data discrepancies cost, GMA's Stegeman suggests that retailers conduct activity-based costing, examining every activity that touches an item or function. "This takes a lot of time and energy, but what you can learn about your operations is invaluable," she says. "This goes beyond IT; it should be a general company issue, as well."
Data synchronization is key in eliminating these types of trading-partner data problems, according to Stegeman. By getting both the retailer and its trading partners to agree on a single version of truth regarding items and processes, a tremendous impact can be made on overall cost and efficiency.
In the GMA-FMI report Trading Partner Alliance Action Plan to Accelerate Trading Partner Electronic Collaboration, the two associations and A.T. Kearney examined the potential benefits of electronic collaboration just by synchronizing basic items in trading-partner catalogs. The research involved case studies that included retailers Wegmans and Shaw's.
As of last March, Rochester, N.Y.-based Wegmans was accepting synchronized item data from a total of 50 suppliers, with a total of more than 2,000 items in sync.
In February 2001 Kraft Foods and West Bridgewater, Mass.-based Shaw's were the first manufacturer and retailer team to complete the UCCnet certification process. In early 2002 Shaw's sent a letter to all its vendors informing them that from Jan. 1, 2003 it would accept item updates only in electronic form -- either via UCCnet or via electronic kiosks at Shaw's merchandising locations. As of March 2003 Shaw's processed 1,637 catalog items and has now synchronized data on 541 of them from a total of 14 suppliers.
Among the retail benefits the study found were:
-3 percent to 5 percent reduction in out-of-stocks. Wegmans realized that 5 percent of its out-of-stocks resulted from supply chain delays due to data integrity issues, such as purchase order processing, freight scheduling, and warehouse and DSD receiving. By eliminating buying and supply chain delays due to catalog errors, data synchronization prevents these out-of-stock incidents, increasing consumer satisfaction and generating several hundred thousand dollars per year in additional sales.
-Two-week reduction in speed to market for new items, which translates to 14 extra days' sales of faster-moving items.
-5,000 to 10,000 hours saved in merchandising and data entry time for new item introductions and updates.
-10,000 to 30,000 hours saved in store labor costs resulting from shelf-tag and scan errors. Shaw's expects to eliminate thousands of hours per year in non-value-added activities caused by shelf-tag and POS errors.
-1,000 to 2,000 hours saved in finance time dealing with invoice disputes related to basic item information. Analysis of the invoice-handling processes at Shaw's revealed that buying and accounting personnel spent hundreds of hours per year resolving invoice disputes caused by basic item-data issues. Data synchronization eliminates these discrepancies and the resulting administrative time spent resolving them.
-1 percent reduction in inventory. Delays in product deliveries caused by item-data errors often cause retailers to maintain a buffer stock of inventory. Shaw's expects to cut current inventory levels once data synchronization has been fully implemented.
There are three steps retailers should take when making the journey toward data synchronization, according to Sterling Commerce's Stelzer. The first step is internal synchronization, which involves eliminating any discrepancies between data to be published on UCCnet and business documents or processing.
The second step, external synchronization, is to ensure that both trading partners begin the synchronization experience in agreement on what the truth of the item data actually is.
Finally, there is ongoing synchronization. This is the process of coordinating all changes pertinent to data synchronization with trading partners, so that they both stay in sync as things change.
"If you've done data sync in a way that will deliver the promised benefits to you and your customer, then you'll be able to establish and maintain consistent, accurate business information values within and between the systems of you and your customer. Whether you achieve it with systems and automation or with boxes and baling wire, that should be your quest. Margins are too thin and competition is too tough to ignore it," Stelzer says.
Judging by the increased number of grocers joining UCCnet, it's definitely not being ignored.