13.1 The Importance of Customer Identification | Internet-Enabled Business Intelligence


Team-Fly

	Internet-Enabled Business Intelligence By William A. Giovinazzo
	Table of Contents

	Chapter 13. Swimming in the Clickstream

13.1 The Importance of Customer Identification

One of the conclusions in Chapter 12 is that the customer-driven organization's objective is to develop a long- term , mutually beneficial relationship with the customer. A key element in this relationship is the a sense of confidence on the part of customers, a feeling that the organization both understands and is able to meet their needs and wants. To understand these needs, we must listen to our customers, establishing a dialog with them.

In Chapter 11, we talked about the neighborhood grocer who provided customized service and developed a customer relationship simply by dealing with people as people. The small shop had a small number of customers. The shop owner knew every customer, and they knew the shop owner. As the economy grew and the small specialty outlet was replaced by the mega-superstore, this understanding of the customer was lost.

Today, brick-and-mortar organizations attempt to reestablish this intimacy with the customer base through technology. Although retail outlets cannot record each and every action of the customers as they move through the store, they do have a record of what customers purchased. They can analyze this information to understand customer behavior. We refer to this as market- basket analysis. In such analyses, we review what the customer actually purchased and combine that with other information to get an insight to our customer behavior.

My family typically shops at the same grocery store every week. We'll use the name Fred's in place of the store's real name . Every time I stop in, I am asked, "Did you bring your Fred's card?" Heaven help me if I say no. Fred has trained his cashiers to be like commandos when it comes to that card. "You really should have it with you, you know. Would you like to apply for one? Please, it will only take a second." Fred is wise in training his staff to be so diligent. The market-basket analysis that little card enables tells him who you are and identifies your buying habits.

Just recently, I stopped by Fred's to pick up a few things on the way home from the office. There were two people in line in front of me. The man being checked out was buying two six- packs , a bag of chips, and some beef jerky. So much for the myth of Southern California's health food culture. As he was about to pay, the cashier dutifully asked, "Do you have a Fred's card?" He committed the sin of all Fred sinshe tried to check out without a card. The cashier, in her exuberance to serve Fred, insisted that he use a Fred's card whether he had one or not. The second man in line was asked if he would mind allowing the other customer to use his Fred's card, which he readily did. The second man was an older gentleman. His purchases consisted of dental adhesive , fresh fruit, some mothballs, and a fiber laxative. As I looked at the second man's purchases, I realized that the cashier had no idea of the purpose behind those cards. Although it probably did not have enough statistical significance to appreciably skew the market-basket analysis, she was putting bad data into the system. To understand why this creates bad data, let's look at how Fred uses the information.

The Fred's card provides Fred with information that is critical in understanding customer behavior. It tracks my purchases for a particular visit as well as each of my visits. It provides continuity. Without the Fred's card, there is no way to associate a visit to Fred's on Saturday with a visit on Wednesday. This matter is further complicated when Fred has more than one outlet. Let's assume that Fred opens a new store near my office. Instead of stopping by the store on the way home, I walk over to the store near my office and pick up what I need. Without the Fred's card, all Fred would see is a reduction in traffic in the store near my home. He would have no idea that the new store was cannibalizing business from the older one. One might suggest that we can link visits by credit card numbers . The problem with this method is that it assumes that I have only one credit card, or at least only one that I use for groceries. What happens if I use cash or a personal check? Could we be satisfied with just capturing purchases made by people with credit cards? People with credit cards may have different buying habits than those without.

The market-basket analysis gives Fred an insight into my buying patterns. It tells Fred that I stopped by twice last week to pick up some scotch. It tells him that this week I picked up two more bottles. If I do it again next week, Fred will be calling Alcoholics Anonymous. Fred stores my purchases and my buying patterns in a central system. He then analyzes those purchases to detect buying patterns. One of the things Fred then does is generate a string of coupons on the back of my receipt. These coupons are based on my buying patterns, directing me to products or services that may be of interest to someone in my particular demographic with my particular buying patterns.

As we noted earlier, the customer-driven organization seeks to develop a mutually beneficial relationship. The card also tells Fred information that profits him. When my data is aggregated with others, it describes the buying patterns of different demographic groups. Fred can use this information for stocking purposes; stores located in large senior citizen communities may be stocked differently than stores in areas with a large percentage of young families. He may elect to stock based on the ethnicity of the community. Perhaps a store in West Los Angeles, where there is a large Jewish community, is stocked differently in September during Yom Kippur than a similar store in the Ventura Valley, which has a smaller Jewish community. Fred provides better service to his customers by tailoring his service to meet their needs while increasing the profitability of his shelf space.

Let's not forget those coupons that are printed on the back of my receipt. These are also of mutual benefit. The selection of coupons was based on my individual buying patterns. If my purchases included veggie burgers and alfalfa sprouts, it would be probably not be a good idea to print up a discount on a jumbo size bag of pork rinds. The selection can take into consideration which products Fred wishes to move or products that are located in a section of the store where he is trying to increase traffic.

Now consider how the cashier who used the wrong Fred's card corrupted the market-basket analysis. We had a senior citizen who was identified as purchasing beer and salty sack food. In addition, the system shows the customer made two trips to the store that day. What would happen if this was done on a consistent basis? Suddenly, Fred would start pushing beef jerky and beer to geriatrics. Not only would the CDC be less than enthusiastic, so would the customers themselves . They would go to Fred's expecting to find dental adhesive and find instead a two-for-one special on beef jerky with every six-pack of Fred's Brand Ale. What good would a half off coupon on salt-vinegar potato chips do a 76-year-old man with pulmonary edema?

Market-basket analysis does have its limitations. It cannot, for example, tell us what a customer put in a basket and then took out. Nor can it tell us the path that the customer traveled through the store. About 7 years ago, I saw a demonstration of a system that attempted to plot a customer's path. The products in the shopping cart were plotted on the store floor plan. The system then plotted the most logical path from one point to the next. This path was just an approximation . It could not tell with any real surety if the products were purchased in any particular order or whether the path projected by the system was accurate in any way. The store hoped that by aggregating this data, it could discover the least trafficked and most heavily trafficked areas in the store.


Team-Fly

Top