Half the money I spend on advertising is wasted. The trouble is, I don't know which half! William H. Lever Founder of Lever Brothers, Ltd.For Jim Dawson, Vice President and MIS Director for Tom Thumb supermarkets, September of 1989 was a month of nervous anticipation. During the previous two years he had worked to breathe life into the Promise Club. Now it was being tested with real customers in three of Cullum's chain of Tom Thumb supermarkets. If the frequent shopper club, electronic coupons, and electronic check worked as well as Dawson hoped, Cullum would soon see gains in both efficiency and market share for the three pilot stores. If these benefits were achieved Dawson would likely face a chainwide roll out starting sometime after the first of the year. The picture after that became less clear. Dawson talked to a visitor about the future:
Our main objective is to sell groceries. The real payback will come when we learn how to use the Promise Club data to help achieve that objective. But for awhile we will be exploring and learning. Initially we saw this as a way to save some time at the checkout and to reduce our exposure through paper checks. We are still looking to achieve those benefits, but they have now taken a back seat to the potential revenue generation possibilities of this program.
Hostile takeovers, mergers, and leveraged buyouts reshaped the grocery industry in the 1980's, and Tom Thumb did not escape attention. By 1988, now a large regional operator of supermarkets and drug stores with related meat packing, distribution, and real estate operations, Cullum Companies had become a desirable acquisition target. On April 27, 1988 Cullum discussed the possibility of a leveraged buyout (LBO) with representatives from the investment bank of Morgan Stanley. Eight months later the firm was taken private by Cullum family members, senior managers of the Cullum organization, and Morgan Stanley. (See organization chart in Exhibit 1). In 1987, the last year in which the firm was required to file financial statements, Cullum Companies had produced slightly over $1 billion dollars in sales and generated $18.1 million in net income. Annual return on net equity from 1985 through 1987 had averaged approximately 17%.
In August of 1989, Cullum Companies owned 53 Supermarkets in Dallas and another six in Austin, Texas. Thirty of these stores included Page drug stores. Cullum also owned fourteen other drug stores and a partial interest in the two Hypermart stores. In an industry with average profit margins of 1% and low growth rates, Tom Thumb faced stiff local competition. The other major grocers in the Dallas market included Skaggs Alpha Beta and Minyard's ( regional competitors, each with about 13% market share in 1988) and Krogers, with a local market share of approximately 20% and 45 stores. In 1988, Krogers had sold 16.3% of the estimated $311 billion in groceries purchased in the U.S [FN1]. A current survey [FN2] showed Tom Thumb to have been the Dallas market leader in 1988 with 26.6% of the local market.
Cheryl Flannigan, an accountant and mother of two young children, was a typical Tom Thumb customer. Cheryl, 31, and her husband had lived in the Dallas suburb of Plano since 1981. During the first week of September, 1989, the Flannigans had received in the day's mail an unsolicited application to join a frequent shopping program being offered by Tom Thumb. After examining the application and marketing literature that accompanied it (see Exhibits 2 & 3.), the Flannigans decided to join the club. Theirs was among the 2,523 application received by Tom Thumb during the first week of the program.
Two weeks later the Flannigans were notified that their Promise Club membership cards were available at their neighborhood Tom Thumb market. An attached note explained how to select a personal identification number (PIN) to ensure the security of the card. They would soon receive the first issue of a monthly newsletter, The Promise Keeper, featuring product specials available only to Promise Club members as well as product information, free recipes, and heath and nutrition advice. In subsequent months, their monthly mailing from Tom Thumb would contain certificates good for store merchandise, based on the total dollar amount the family's purchased in the previous month.
The following day Cheryl picked up her Promise Card at the neighborhood Tom Thumb. Using one of the three Promise Terminals located in the store's cash office she established the family's PIN number. This number would remain associated with the card unless she later requested a change - at the same store. An hour later, after selecting her groceries, Cheryl rolled her shopping cart up to the check-out counter. With a little assistance from the check-out person, she "swiped" her membership card through the Promise Terminal which read her membership number from the magnetic strip embedded in her card. The number was instantaneously transmitted to one of the two IBM AT personal computers located in the store's business office. In less than two seconds, the computer determined that Cheryl was a Promise Club member and that she was authorized to use either a preapproved personal check or a new electronic check to pay for her groceries. She selected the button next to the electronic check option, entered her PIN number, and then put her membership card back in her purse.
As the clerk scanned the grocery items, the point-of-sale system automatically checked for membership specials, crediting Cheryl with the Promise Savings for appropriate items. As the items were accumulated, their universal product code (UPC) identifier, quantity, and price were transmitted to the store-controller computer in the store office. The computer appended the time of day to the item data and then stored all of it on an attached hard disk. As Cheryl wheeled the grocery cart out of the store, the total amount of her purchases for the day was transmitted to the store-controller computer. Here the amount was added to a running total of the Flannigan's purchases for the month that was stored in a "household" file. At the end of the month the Flannigans would receive certificates redeemable for merchandise based on the total of their purchases for the month. If Cheryl shopped in another participating Tom Thumb store, the total value of those purchases would be automatically forwarded to her home store computer.
We all failed miserably with these systems. Transaction volumes often represented less than 1% of total customers. First, the stores only accepted cards from one major bank. Second, the customers often thought they were losing a couple of days of float if they used the debit card. In reality, the paper checks got processed in less than 24 hours, but what the customer believes is more important than the reality. Finally, many customers assumed that there is a fee associated with using a card to obtain money.
Dawson, who had joined Cullum in 1974 as a systems analyst, kept searching for new solutions. In 1985, he had become aware of Amarraca, a grocery store in Pittsburgh , which had implemented a delayed debit card. The delayed debit card, like the personal paper check it was designed to replace, did not instantly transfer cash from the customer's bank account. Instead the debit card transactions were accumulated and, at the beginning of the next business day, transmitted to the bank for payment. The customers perception of "float,"was preserved [FN3] . Amarraca's debit card had achieved the highest transaction volumes of any debit card being used in the country. With the encouragement of Jack Evans Jr., Tom Thumb's president, Jim Dawson decided to look more closely at EFT using a delayed debit card.
In November of 1986, Dawson was approached by representatives of Citicorp POS Information Services Incorporated (CPOS). CPOS had been founded in the mid 1980's with the objective of developing the National Purchase Behavior Data Base. Through agreements with major grocery chains, the firm sought to capture consumer purchase data at the point of sale, buy that data from the retailer, and then sell the repackaged data to consumer package goods manufacturers. In February of 1987 CPOS had helped Ukrops, (a Richmond, Virginia based grocery chain) pilot test an electronic couponing system. Members received a monthly newsletter from the store and were then automatically given a discount off selected items at the check-out stand if they presented their bar-coded membership cards. The Ukrops "Coupon Bank" pilot was successful and, by October of 1987, the system was rolled out across Ukrops 19 stores. Jim Ukrops, president and chief executive officer of Ukrops, described the system benefits in May of 1989 at the Food Marketing Institute's annual convention:
Over 180,000 households are already members of Ukrop's valued customer program, and we're signing up about 600 more customers per week...After the first year of the program chainwide, our identical store sales grew 10% and our market share grew from 25% to 30%...In zip codes surrounding our stores, up to 80% of the households have the cards...Sixty percent of all store transactions - representing just under 85% of total sales volume - are made with this card4 [FN4] .
But in early 1987 Jim Dawson had his eyes set on the efficiency benefits that he felt could come from a delayed debit card and the benefits at Ukrops were still in the future. Moreover, Tom Thumb was already involved with an electronic couponing program offered by another vendor and, at the time, CPOS's had little to offer beyond their Coupon Bank. CPOS, however, was still anxious to get access to the Dallas market and to develop new information-based products. They agreed to support Tom Thumb's efforts to introduce a proprietary debit card. An agreement was reached with Cullum in March of 1987, and the first project meeting was held in October of the same year. The proposed delayed debit card was presented to customer focus groups seven months later. Jim Dawson describes the focus groups' reaction to the proposal:
We were very disappointed. They told us, 'We don't want your blankety-blank cards!' The message they sent us was, 'What's in it for me?' We just weren't adding enough value to the customer.
Diane Vaughan, a Vice President at CPOS, was assigned to the Tom Thumb project as the Client Relationship Manager in January of 1988. She had previously been employed in another division of Citicorp where she had worked on EFT projects for the supermarket industry. She commented on the Focus group experience:
I wasn't surprised that the focus group participants didn't like the idea of the debit card. EFT has proven to be very popular with customers who use it, but it is difficult to visualize. After the focus group experience, Jim Dawson was convinced that the project needed to be marketing driven.
In the intervening months, Dawson and other personnel at Cullum attended several conferences on frequent shopper programs. That, and the now apparent success of the Ukrop's program, convinced them to take a closer look at buyer incentive programs. Diane Vaughan described one important convert:
Initially Jim Dawson was looking for operating savings and increased throughput with EFT. Tom Arledge [Senior Vice President of Tom Thumb operations] was not very excited about the project when it just focused on operations. But when the marketing implications became apparent, he quickly came on board. His support was critical to the success of the project.
In July of 1989, Arledge was anticipating a 10% increase in sales from the Promise Club, but he was also interested in solving a customer service problem. As he described, customer identification had at times been an embarrassing problem:
For some years we kept facing the same question - how can we establish a process to identify our good customers when they come to cash a check? Women in particular got offended because we were constantly asking them to show their driver's license. Jack [Evans] Jr. gave Jim Dawson and me the challenge to solve that problem.
In the spring of 1988, Cullum and CPOS prepared to present the Promise Club program to customer focus groups. This time, the proposed program, which now included buyer incentives as well as the delayed debit card, received high marks. Cullum management budgeted the money necessary to go forward with a pilot implementation. CPOS agreed to cover the cost of the communications software development and marketing. Beginning in the summer of 1988, however, the germinating leveraged buy out of Cullum put brakes on the project. Tom Arledge explained the decision to put the Promise Club project on hold:
The grocery business has historically been very conservative and so have we. From the beginning of my career to the present day, our challenge is to run the store like you own it. In the past five years our industry has experienced tremendous change and has become a very fast paced innovative business. Historically, we at Tom Thumb-Page have not been the pioneer of new ideas but have prided ourselves in taking new ideas and bringing them to a higher level of perfection.
At the time of the LBO, it just didn't seem like we had either the resources nor the management time to be out front on this one. We were attempting to cut 10-15% from our headquarters expenses and we decided that if we couldn't do this right, then we shouldn't do it at all.
By Thanksgiving of 1988 the Promise Card project was nearly dead. CPOS had working agreements with nine supermarket chains in January of 1989, but was anxious to have access to information on the Dallas grocery market. In a January meeting among Cullum and CPOS senior management, the two firms negotiated a modification to their original handshake agreement that would permit Cullum to go forward with the pilot tests without adversely impacting their financial position.
In January Dawson and CPOS personnel estimated that the pilot could be launched in the late spring of 1989. Because of various delays, including the necessity of having a specially programmed EPROM [FN5] card for the VeriFone terminal, the pilot was not launched until August of 1989 with "friendly" users (store personnel and their families) and a month later with normal customers of the three stores selected for the pilot - stores 45, 54, and 68.
Each of the stores in the pilot program were located in Plano, a suburb north of Dallas. Frito-Lay, a producer of salty snacks, told Cullum that the three stores represented 40% of their products' sales in the Plano market. But there was little reason for complacency; in the first quarter of 1990 Kroger was scheduled to open a new super store just across the street from Store 68. The pilot stores varied considerably in the type of customer served Store 68's customers had an average household income of nearly $50,000. Comparable figures for Store 54 and 45 were $35,000 and $20,000 respectively. In July of 1989 Burton Hubbard, store director of store 68, sat with a visitor in the restaurant in the front of his store. As he talked he stopped frequently to greet, and occasionally to help, customers. In quieter moments he described the differences among the three stores included in the pilot:
Store 54 is just 3 miles away from here but, given its demographics, it could be in a different state. Store 45, seven miles further down the road, might be in a different country - it is so different from us. Of the three stores, Store 54 has the highest percentage of coupon redeemers, but I think the Promise Club will be different. I think I can sign up 50% of my customers within the first six months and they will stick with it. My customers are among our most loyal. People will see this as a status symbol, and it will be convenient. It will help them organize their use of coupons.
Hubbard was enthusiastic about the Promise card. In July of 1989, a month prior to the pilot test, he summarized his expectations:
This will speed up the paying process and I'm in favor of anything that gets people through the line faster. But we will see real gains from advertising. We are spending a fair amount of money on plastic [membership cards] but we will recoup that with savings in advertising. Once my customers have this card we'll stop shooting in the dark. If you live in Oak Cliff you don't want to hear on the television or read in the newspaper that chickens are on sale in Plano. With the Promise Card Mr. Cooper, my zone director, and I can make the decision to flyer up an area with our circulars.
Hubbard had become involved in the Promise Card long before the pilot project. He and the other pilot managers served on a project development committee that met each week. The store managers were invited to the meetings on an occasional basis. Hubbard explained:
They called us in for the meetings but not all the time. We represent the real world. We tell them what's going to work with the customers and what isn't. They want to hear some of that, but computer people prefer to talk to other computer people. Their job gets tougher when they come in contact with the real world."
This won't work if you fill the newsletter with coupons worth 5 cents off on a bag of kitty litter. The discounts have got to be substantial and there have to be enough to fill the newsletter. Joe Melton could immediately see the potential of this and will make it very profitable for Cullum. Joe is a wizard at dealing with the packaged goods manufacturers. In May he filled the Cullum cafeteria with manufacturers representatives and sold them the program.
Melton felt that his role in the Promise card had been to sell the vendors on the instant savings part of the program.
The Program doesn't really have much to do with me. This is really a high tech project coming from Jim, "high tech", Dawson and his data processing people. My role was to introduce the program to the suppliers. We invited 200 of the brokers and manufacturers representatives in for a meeting to present and discuss the Promise Program in detail. The brokers and manufacturers were pretty skeptical. We already have nearly 3,000 green special price tags hanging from our shelves plus store coupons listed in the papers. To them this was just another program on top of what they were already doing with us. After the meeting they didn't call us and sign up as quickly as we thought they would, most of the vendors had to be called. Most of the vendors wanted their marketing people to spend funds on this rather than take it out of their own sales promotion funds. We told them that this would increase their sales by 20% and they got nervous. At that time most of them felt the program would be too expensive if the redemption produced a 20% increase.
A manufacturer is interested in only one thing - moving cases of merchandise. Market share is what they look at. I think the Promise Card Club will help us increase share and, if Dawson can get sales-by-item information to us on a timely basis then it will be easy for us to sell the program in the future.
The computer and communications hardware and software required to support the Promise card required approximately $18,500 in additional equipment in each store in addition to the scanners and store computers already installed. As shown in Exhibit 4., each store was equipped with two IBM PC AT personal computers and an IBM Model 25 personal computer. The Model 25, along with three Promise Terminals, was located in the store's cash office. The first of the store's two PC AT computers, called the PC controller, included 130 million characters of hard disk storage and 2.5 million characters of main memory. The PC controller and the Model 25 were each connected by a "store loop" communications link to the store's existing IBM point-of-sale (POS) terminals. These, in turn, were directly connected to a bar code scanner supplied by IBM, NCR, or Spectraphysics. The store loop was capable of transmitting at either 4,800 or 38,400 bits per second and was used to provide a price lookup capability for the 3683 or 4683 POS cash registers. As customer purchases were scanned, the POS terminal used the store loop to transfer back to the PC controller the so called "session data." The session data included identifying information on every item sold to the particular customer. A local area network (LAN) connected all of the store's Promise Terminals to the Verifone 1800 membership authorization devices. If a customer wanted to cash an electronic check, the Promise Terminal sent a message over the LAN to the Verifone 1800 which passed the request to the PC Controller for validation.
Check cashing authorization was approved by the PC Controller in the customer's primary shopping (home) store. If a customer visited another store the authorization request was forwarded out to the Payment Authorization Switch (PAS), which electronically routed the request to the customer's home store for approval. During the course of a day the electronic checks were transmitted to the PAS. Each morning, after balancing the PAS with a total received from each store, the electronic checks were transmitted to CitiBank in Delaware. CitiBank Delaware acted, for a fee, as Cullum's automated check clearing house for processing the electronic checks.
The two store PCs were connected by an IBM Token Ring Local Area Network, capable of transmitting data at 4 million bits per second. The second computer, which included a printer, served as a backup to the PC controller. It was also used to transmit the daily item transaction logs each night to CPOs facility in Stamford, Connecticut. CPOS, which purchased the data from Cullum, provided a separate, 19,200 bit per second, modem and dial-up line to collect this accumulated session data on every item sold by the store that day. The token ring network linking the stores two personal computers permitted the session data to be transferred in four or five minutes between the two personal computers prior to its being picked up from the administrative computer by CPOS. It had become necessary to use the second computer for transmitting to CPOS because of a lack of hardware slots [FN6] on the PC controller and the additional storage required by the communications software. Each week the demographic data on member families, gathered from their membership applications, was sent by tape to CPOS headquarters where it could be matched up with the session data.
Cullum headquarters was equipped with an IBM 4381 mainframe computer which allowed access to 30 gigabytes [FN7] of direct access disk storage . A separate IBM 3725 communications processor handled communications protocol with the store computer. Electronic check and authorization transactions, however, were intercepted by the PAS. Price changes could be sent down as required from Cullum headquarters or accounting data sent up from the store including sales by item for each day and the accumulated monthly shopping totals of Promise Card member families. Prior to the Promise Card, the store computers communicated to Cullum headquarters via dial up phone lines but, the EFT aspects of the program required that each store be constantly connected to headquarters. This was done with leased lines capable of transmitting at 9,600 bits per second. These "multidrop" lines each supported up to five stores. In the future there was a possibility that the company would move to some sort of point-to-point [FN8] communications technology.
The software required for the Promise Card program was primarily provided by Citicorp. The different models of POS terminals and scanners used by Citicorp's partners compounded the complexity of this activity as did the unique features requested by the different chains. Much of the software had to be built or substantially modified for each retailer. According to Diane Vaughan, Tom Thumb presented a particularly challenging development task for the staff of CPOS employees assigned to systems development:
Cullum is the most complicated environment of the stores that we are working with. We had to develop software that interacts with the bar code scanner, the cash register [POS terminal], the mag stripe reader [the Promise Terminal], and the communications switch that processes the electronic checks. For our other installations we are only interfacing with the cash register and the scanner.
Dawson and Vaughan had set three objectives for the three month pilot study: 15,000 households enrolled, 12% of the customer payments handled electronically, and a 7% increase in store sales. By the third week of October, 1989 some six thousand customers had been asked to pick up their Promise Club cards and, that week many customers would begin to use them. Over sixty percent of these initial applications were approved to use the electronic checking feature - with a somewhat higher percentage in store 68 than in the other two. Some logistical problems were encountered during the first days as numerous customers showed up to pick up their cards at the same time. Each had to be carefully shown how to select a PIN number and this took time. But, according to Jim Dawson, those were "the kinds of things you learn about from pilot studies." Tests the previous month with friendly users had demonstrated that the part of the check-out transaction devoted to paying for the groceries had decreased from ninety to ten seconds. Dawson was confident that as the early adopters were seen using the card, and quickly passing through the checkout counters, that they would quickly sell the program to much of the rest of the customer base.
If the pilot was successful, Dawson thought the system could be rolled out in the remaining grocery stores in as little as six months. The Page Drug stores would require longer as many were not yet equipped with scanning equipment.