You are here
If there is one predictable thing about information systems executives, it is that they almost never, ever throw anything away. This is sometimes regrettable, even if the cost of saving data isn't going to have a major impact on the company's bottom line. Consumer purchase data is saved, but rarely used. While the least of evils is to save anything that just might be used, it means the data is never evaluated for quality or completeness. This month I'd like to outline the issues that will compromise your biggest database, frequent shoppers.
My readers may think that the use of frequent shopper data is an overworked topic. I would argue that its relevance was never more important. In the Sept. 1 issue of Progressive Grocer there was an important article titled "The Age of Real Revenue" written by two executives from Cannondale Associates who shared their firm's 2002 report on trade promotion spending and merchandising. The crux of the report was that changes in the rules for accounting for promotional funds will force consumer packaged goods companies to reevaluate the effectiveness of these expenditures and to improve their productivity.
Promotional expenditures must be treated as a reduction of sales under the new rules, and this reduction must be reflected in a restatement of sales for the last 10 years. Cannondale believes this will translate into reductions in 2002 revenue that will average 8.5 percent.
Many, if not all, retailers depend on promotional monies to fund their merchandising programs. To sustain their merchandising programs in this new world, retailers will have to develop ways to improve the productivity of these funds. The CPG companies believe that they will be moving steadily toward the allocation of promotional funds based on the profitability of the investment with each retailer. The obvious approach, the Cannondale report concludes, will be for retailers to allow the CPG companies to get closer to their customers through data sharing. Therefore, creating quality data will become a corporate priority.
Let's look at the obvious area of vulnerability, the perishable data that represents purchases of variable weight product. Inevitably, this data is incomplete because it was never linked to an accurate price per pound that applied to that specific UPC in that specific store on that day. Why is that so critical? In the analysis of scan data you always need to know how many units were sold. Without the ability to convert the sales to pounds, we have data that isn't viable for analysis.
In tracking data it is incredibly important to understand that most homes have multiple shoppers and if you can't link data from multiple card holders, your analysis is a waste of energy. Any conclusions are incorrect. For example, frequent shopper data may indicate that a card holder never buys meat, while the reality is that another member of the household buys the meat. When most frequent shopper programs were established, retailers were reluctant to ask important questions about their shoppers. As a result, the data is next to worthless. Any attempt to market to these families will always be based on a one-sided perspective of the shoppers and their needs.
A partial fix
In many situations, this issue can be partially resolved by comparing the applications of all your shoppers by last name and address and then linking those with a match. However, when there are large apartment buildings in the area, it is likely you will get matches on both name and address that do not represent a real family unit. I cannot overemphasize how important it will be to your company's future ability to market to your customers.
There are other issues that need to be addressed to improve the data that relates to the products you sell. Tracking customers' purchasing patterns means that you have to understand when they change brands by choice or by necessity. To do this, you must be able to create product linkages between an old UPC and the new one on the replacement product. You need to be able to distinguish a discontinued product without any replacement from one that has a "new and improved" alternative.
It is also important to be able to identify product characteristics, such as sugar-free, salt-free, etc. While some brand names, as they are represented in your files, may give a clue, it is not useful in this format. You need coding associated with each product that can identify these characteristics so that you can extract data efficiently from your data warehouse.
For example, a CPG company wants to promote a new product to all of your customers who are diabetics, based upon their regular purchase of three or more diabetic products. To secure the promotional funds, your company will need to be able to identify the number of customers who match this criterion, deliver the targeted promotion to them, and track and report the results.
You need to be able to recognize when a brand was on sale, and to qualify that sale as to what promotional support or causal data was associated with the item during the promotional period. This promotional coding is critical when you are searching such a huge database. It is not possible to look at each item to determine whether a "hot price" indicating a promotion was in effect.
Causal data can also be tricky if you want to be extremely accurate. Since most retailers have both a must display list and an optional display list, totally accurate data will require that each store provide identification of the products it displays. While the effort required to capture this data is minor; it should provide the store manager with important historical reporting on the productivity of his end-displays.
It will also be good to be able to establish linkage among related products such as hot dogs, hot dog buns, and mustard. Such linkages are very helpful as you add new services to enhance the shopping experience in your store.
Many retailers offer the ability to create a shopping list on their Web sites. Some are now moving to provide purchase history to assist consumers in developing branded shopping lists that anticipate demand by analyzing purchases over time. This analysis can easily "predict" the need to buy staples such as milk, bread, and coffee, and indicate if the customer tends to be a brand switcher on any of these products.
This entire process must be done in a timely manner. Timely means establishing the opportunity within a couple of days and reporting the results during the week after the promotional period has ended. The tools to support this speed of retrieval and analysis may not be part of your existing database support package. That means you have an extended time period to evaluate various database tool sets and select the one that works best for you. This technical priority should be moved to the top of your systems programming task list. You should also consider how you would move a subset of this data to create a file that your CPG partners could access.
I have avoided the issue of how to deliver customer-specific merchandising messages. There are a number of current alternatives. Of course, the most obvious is the Internet. If you have a Web site, you can use it to identify customers and present them with these targeted promotions. The promotions can be queued at the customers' home stores as coupons in the POS system for delivery to any household member—if you can link card numbers to a household.
A kiosk can be used at the store for the same purpose. Marsh Supermarkets recently introduced the MyMarsh Interactive Shopping System in selected stores. It uses a touchscreen kiosk at the POS to deliver personalized incentives in addition to those offered to all customers through the loyalty card program. While Marsh has not released results, you can expect other retailers to try this approach.
I am confident that during the next six months you will begin to read of new technological delivery mechanisms that are being evaluated and even rolled out throughout the United States and Europe. These new offerings will be integrated with the consumer's shopping experience and with a number of interrelated facilities offered throughout the store.
Terminals will be carried by customers or attached to shopping carts. These consumer terminals will allow the retailer to target promotions and provide new services such as electronic shopping lists and kiosks that will take orders. They will also expedite checkout.
These facilities will be bundled into a single, integrated experience to encourage all customers to use them. Other in-store technologies require use by a relatively small proportion of customers to achieve "success." To achieve good productivity of promotional funds, the CPG companies need to know that they can reach the majority of the shoppers in the store who meet their objectives. That means that to achieve a productive environment the technology must become ubiquitous with shopping in your store. This future is around the corner. Will you be ready?