Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Custodial Package Tracking at Federal Express CARL NEHLS Federal Express is a true American success story. Started a mere 15 years ago, it created the door-to-door, overnight express package delivery business and forever changed the way the United States and the world do business. Founded by Fred Smith in 1973 with a fleet of 14 Falcon jets, Federal Express had grown, by 1987, into an international company with $3.2 billion in yearly revenues and more than 50,000 employees worldwide. Federal Express picks up and delivers more than 850,000 packages each day using a fleet of 60 Boeing 727s, 20 DC-10s, and approximately 100 Cessna 208 aircraft. Federal Express's phenomenal growth, and the resulting articles, business seminars, and case studies have made Fred Smith a famous man. However, a frequently overlooked "key element" to Federal Express's success was Fred Smith's realization that being able to provide information about a pack- age's location, status, and movement was just as important as actually picking up and delivering the package. If the package is so important to the customer that he or she needs it delivered overnight, the customer is almost certain to want to know that everything has gone according to plan. Federal Express's information management and positive package tracking systems were started early and quickly grew in importance as thousands of people called to have their packages picked up and many called again the next day to ask, "Has my package been delivered?" COSMOS IIB is the latest in a series of information system developments that Federal Express has undertaken to add value to its product while simultaneously improving its operations. 57
58 CARL NEHLS BACKGROUND The core of Federal Express's information system is COSMOS (Custom- ers, Operations, and Services Master On-line System). Originally developed in 1977, the system was rewritten to run on IBM computers in 1979. These mainframe computers located in the Memphis, Tennessee, corporate head- quarters were used to store customer information and rate tables. Call Center agents accessed this information when customers called to request a pickup, and the requests were sent to a printer in the appropriate station. Initially, couriers used pay telephones to call in and receive their dispatches; however, radios were soon installed in the Federal Express vans for quicker and easier . . communications. During the next several years many more functions were added to COS- MOS to facilitate the collection and distribution of customer, ZIP Code, and operations information. By the late 1970s software applications totaling nearly 1 million lines of code were running on COSMOS. The network, comprising a cluster of IBM 3090-class machines with thousands of terminals located throughout the world, continues to be one of the largest IBM networks in existence. Additional computer systems such as FAMIS (Field Activities Manage- ment Information System) and ORBIT (On-line Revenue Billing & Invoicing Technology) were developed to handle the company's payroll reporting, field reporting, and billing functions. Although these systems received limited amounts of information from COSMOS, each system remained essentially independent. Some standard information, such as ZIP Code and location data, was maintained separately on each system, and in some cases signif- icantly different conventions were followed. For example, COSMOS kept track of stations by the three-letter abbreviations for the local airport (MIA for Miami, Florida), whereas FAMIS used a number to identify a location. Keeping these multiple systems synchronized was a constant problem and would later pose a systems integration problem for COSMOS JIB. Providing information to customers about their packages was becoming a significant challenge as the package volume grew. Exception packages- those held for pickup by the recipient, those delayed because of weather or other problems had information key entered into COSMOS. However, if a customer wished to know the status of a nonexception package, the station had to be contacted by telephone or printer message, at which time local employees researched the package and phoned the information back. By 1977 bar-code technology was emerging as a reliable way of capturing information in a very efficient manner. Federal Express began to investigate the idea of bar coding the package number on each airbill. The concept was to record the number of each package each time it changed hands. If that
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS 59 information was provided to COSMOS, a customer could be told the exact status of his or her package while on the telephone. To further improve the tracking system, additional information could be collected for each package at certain points. For example, as the package is being delivered, the name of the person signing for the package could be entered. By early 1979 Federal Express was outstripping the supply of technical people in Memphis and elected to construct a Technology Center in Colorado Springs to help attract talented people to the company. One of the first people in Colorado Springs was Harry Dalton, who had spent his early days with the company as a station manager in Philadelphia and later as a district director. Dalton, then a director reporting directly to the president, was sent to Colorado Springs and matrixed to the Advanced Systems Group (that is, he was assigned to the group, but he did not report through the group's chain of command). Dalton explained his unusual placement in Colorado Springs as a desire of executive management to "inject a dose of reality into the bits and bytes group." In mid-1979 Dalton was given the go-ahead to conduct a field test of the available bar-code scanning equipment. Federal Express's first survey of the hand-held terminal technology market revealed a small selection of available hardware. Most of the devices had very limited keyboards for data entry and were capable of scanning just a few types of bar code. Norand Corporation was just preparing to release its first programmable terminal with a numeric keypad (see Figure 11. Federal Express acquired some of the first terminals Norand could make available and began testing these models as well as terminals from other vendors. The plan, dubbed COSMOS II, was to supply each courier with a scanning device for use in the station and on the road. The important tests were to determine if hand-held devices were durable enough to use on-road, to see whether large quantities of sequential numbers could be printed in bar code, to determine the impact of scanning on field people's morale, and to provide data that could be used to predict the impact of package scanning on the entire company. The scanners would be used to scan the package-tracking number of each package every time it changed hands from the point of pickup to the point of delivery. The information collected by these scans would be transmitted from bases in the station to COSMOS, where a customer service agent could access the information from a terminal. These. field tests showed that scanning the packages as they changed hands was a feasible plan; however, the equipment was not sturdy enough to with- stand the rigors of a delivery route. As a compromise the project was broken into two parts, COSMOS IIA and COSMOS JIB. COSMOS IIA would only involve scanning packages in the stations where the equipment could survive. More durable equipment had to be found before the on-road scans of COS- MOS IIB could be implemented. This type of compromise not allowing
60 CARL NEHLS FIGURE 1 Norand 101 XL. "best" to get in the way of "better" was to become a common approach to many technical challenges at Federal Express. The equipment selected for the in-station scans was a Norand device that was modified to incorporate an alphabetic keyboard. This special device, later adopted as a Norand standard, was referred to as an HHT, or hand- held terminal (see Figure 21. The software for this unit was specified by Federal Express, written by Norand, and stored in PROMs (Programmable Read Only Memory). In early 1981 COSMOS IIA implementation was begun. A group of six people crisscrossed the country installing 450 HHTs and training personnel in 150 stations on how to scan the 100,000 nightly packages. By the summer of 1981 Dalton had moved to a director's position responsible for Corporate Industrial Engineering in Memphis, and began leveraging technology appli- cations through this department. Eventually, most of the work with bar codes
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS 61 a. ~ ~ FIGURE 2 Norand 101XL Alpha 1 (HHT). ~ . and scanning equipment was shifted to others in the group. Federal Express engineers, particularly David Dietzel and Jim Turpin, continued to watch for and test new scanning equipment as it hit the market, searching for the device that could support COSMOS JIB. The reaction of station management to the implementation of COSMOS IIA was often less than favorable. Extra personnel were required to perform the scanning, and station personnel thought that the scanning hampered the speed of the morning operations. After several months of operations it became apparent many stations were not scanning as diligently as had been hoped. Soon compliance reports comparing the number of scans done by each station to their package delivery counts were developed. Not long after that, the stations' scan compliance was included in the MBO (Management by Ob- jective) system for the station managers, thus supplying even more incentive to make sure the scans were done properly.
62 CARL NEHLS Not all the plans for COSMOS IIA were made perfectly. Some of the scans in the HHT were found to be impractical, and the need for new scans soon developed. It was frustrating to Dalton's engineers that the software in the HHTs could not be easily adjusted to meet the changing needs of the field operations. Software changes to the HHT were difficult because not only did Norand have to be contracted to make the changes, but there were also significant logistical problems in locating all the terminals and perform- ing a PROM swap without disrupting operations. Despite these problems and frustrations, COSMOS IIA scanning became a natural part of station operations and the core of Federal Express's package tracking and tracing operations. The lessons learned in COSMOS IIA would play a major role in the development of COSMOS JIB. By 1979 daily pickup operations in some cities had increased to the point that 2 to 3 dispatchers were issuing nearly 3,000 requests for pickups to more than 100 couriers in the larger locations. Many of these dispatches occurred just before local businesses began to close, which made the dispatch function almost impossible to manage. Because radio frequencies were tightly reg- ulated and in short supply, it was impractical to use multiple radio channels. Instead, Federal Express went to the marketplace and had technology created in the form of DADS (Digitally Assisted Dispatch System). The first DADS system was implemented in mid-1982. DADS was developed with a Canadian company, Mobile Data Interna- tional, and involved having COSMOS send the request for pickup to a computer in the station instead of to a printer. The dispatcher would confirm which route should receive the request and would then forward the request to a computer terminal mounted in the van (see Figure 31. Sending these short bursts of data instead of conversation on the radio dramatically reduced the load on the radio channel and provided a real time link between the courier and Federal Express's computer system. The ability to pass digitized data between the van and COSMOS opened up an entirely new field in data communications. Dalton and Dietzel rec- ognized that this real-time link to the computer system would allow scan information to be immediately transmitted to COSMOS rather than waiting until the courier returned to the station. Making the information available to the customer service agent in a near real-time fashion would eliminate even more traces. For example, a trace request at 10:00 A.M. without DADS would still require a call to the station, because couriers would still be on their delivery routes. With the real-time communications of DADS available, the information of a 9:55 A.M. delivery could be provided at 10:00 A.M. and prevent the call to the station. In the early months of 1982 efforts intensified to find a COSMOS IIB device. By late 1982 it had become quite clear that the manufacturers of bar- code scanning equipment did not have plans to develop what the engineers
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS ~ Aft:..: l ~.~....~ ~ ~ . ., ~ . ......... . . i...] 1 1 ..., !! FIGURE 3 DADS Mobile terminal. The "shoe" for the scanner, shown in this photograph, was added after DADS implementation. 63 of Corporate Industrial Engineering were looking for. Consequently, a Re- quest for Proposal (REP) was issued in the spring of 1983 for a device with no keyboard and 4K of RAM (Random Access Memory). Five thousand devices were to be produced, and they were to be accompanied by charging and communications boots. Six companies responded to the REP. Most of the companies, already established in the scanning equipment industry, re- sponded with proposals that used part of their current line of equipment. One exception to this was a capable start-up company in North Carolina named Hand Held Products (HHP). During one of the REP review sessions with HHP, they were told that they must bring back exactly what was requested. Turpin held up a large felt-tipped pen and said the scanner must "look and feel like this." Turpin explained to HHP that (1) the device must read Codabar, (2) the device must
64 CARL NEHLS fit in a shirt pocket, and (3) HHP must develop a unit for presentation to Harry Dalton and it must work. Representatives from HHP visited Memphis 3 months after the REP went out and had with them a working prototype. This device was nicknamed "tent peg," because of its shape. HHP had been a dark horse, but suddenly the company had made itself a front runner. COSMOS IIB 1974-1983 Summary Harry Dalton and his engineers had finally found a vendor that could and would provide a scanning device the courier could carry in his or her pocket and use to scan packages on the road. The compromise solution of only scanning packages as.they arrived at and departed from the station had become standard operating procedure. The ultimate goal, first envisioned in the mid- 1970s, of collecting tracking information at each change of possession was still there. By 1983, however, the objectives of COSMOS IIB had been expanded. Not only was tracking required every step of the way, but the information was desired instantly and had to be managed to identify packages that did not appear at the next checkpoint. The DADS system was to be used to transmit the data to COSMOS so that the information was available to customer service agents within seconds. Computers in the stations would generate reports of packages that did not appear at the correct destination, thereby enabling the packages to be recovered and delivered on time. initial Financing Harry Dalton had been selling the project on and off over the years, always looking for ways to enhance the project's value. Many of the benefits were related to the courier force that would be using the device, but Dalton also canvassed the rest of the company to identify other benefits. Marketing, sales, and customer service departments identified savings and profit potential that significantly added to the benefits that could be obtained from labor savings, but most of the benefits still were not considered solid enough for inclusion in a financial justification. Finally, after reviewing the entire project plan in May 1984, senior management saw enough benefits to approve de- velopment and test funds. Preliminary implementation plans called for 13,000 scanners to be rolled out by June 1985. Initial Hardware Development The Memphis-based engineering group spent quite a lot of time working with HHP to develop the ultimate scanning device while a search for a station
CUSTODIAL PACKAGE TRACKII!iG AT FEDERAL EXPRESS 65 FIGURE 4 Hand Held Products Prototype II. processor for generating COSMOS IIB reports was centered in Colorado Springs. Dealing with a start-up company such as Hand Held Products brought with it a unique set of circumstances. As a small, aspiring company it carried the vigor of youth, and the energies its employees poured out produced speedy results and changes. The requirements for the COSMOS IIB device quickly evolved from a simple "tent-peg" scanner to a scanner with a display (see Figure 4), and then to a more advanced unit with the ability to make some elementary keyboard entries (see Figure 5~. Working prototypes of all these units were produced by HHP in less than 6 months. By July 1984 a FIGURE 5 Micro-Wand I (Hand Held Products).
66 If........... : ""....... ~ ~ ~ I. CARL NEHLS . . . :::::::: :::::: :::::::: .~ ....................... ... .... ............. .. .. . ... ~ . ~ ~ ~ . ~ ~ ~ ~ ~ ~ . . . . I. FIGURE 6 Supertracker (Hand Held Products Micro-Wand II), a subsequent model of the Tracker. decision was made to manufacture a scanner with a full alphanumeric key- board, a two-line display, and 64K of memory (see Figure 61. Federal Express eventually purchased 250 of these units for field testing. The development and production of these units were expected to take nearly a year. Although developing a new piece of equipment had its advantages, it also had its disadvantages. For example, part of the unique design for the Federal Express unit, appropriately named Tracker, was an 8-bit surface-mount pro- cessor that had just recently been introduced. The COSMOS IIB project required some of the first batch of these chips to meet its development schedule. Special arrangements had to be made with the manufacturer to secure the order to ensure that the development schedule could be maintained. In Colorado Springs, bench tests for the station processor had been de- veloped and computers from about a dozen vendors were evaluated. Each station was to receive a processor. All the scan data would be fed through this unit either through DADS, a communications and charging base in the station, or modems connected to telephones that the couriers could call when transmitting their data acoustically. The data would be stored locally for report generation and then, if appropriate, sent to COSMOS for use by the customer service agents. After much evaluation, the newly developed MicroVAX from Digital Equipment Corporation was selected based on its functionality and modularity. David Dietzel was named manager, and Jim Turpin was also promoted at this time and managed a hardware design and development group that became instrumental in the project's success. Initial Software Development With the hardware specification complete in July 1984, the staff in Col- orado Springs began to write the software for the Tracker, the Station Pro
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS 67 cessor, and COSMOS that would make the system work. Because no production Trackers were available to develop software on, a desktop computer was connected to a prototype Micro-Wand I and used to emulate a Tracker. The initial programs for the Tracker were very straightforward; the courier was forced to answer the same set of questions no matter what type of situation was being documented; no decisional branching was provided. All the soft- ware for the Tracker was written in assembler and stored in EPROMs (Eras- able Programmable Read Only Memory. The first version of software covering a total of 10 different types of scans required about 24K of the 32K EPROM. The data actually collected by the courier were stored in 32K of RAM. Programs to extract the information from the Trackers and produce reports were written for the station processor. These programs were written based on the specifications for the Tracker software, but none of the programs could be tested in anything close to a production environment until the first hand-built Trackers were provided in March 1985. Project Management Through the end of 1984 all project work for COSMOS IIB had resided entirely within the departments of Harry Dalton and Jim Tollefson, a director in Colorado Springs. As the dates for field testing and, hopefully, imple- mentation grew near, it was apparent that COSMOS IIB would require the involvement of nearly every area of Federal Express to make it a successful project. A meeting was called in January 1985 and representatives from every department in the company were invited; nearly 150 people attended. A briefing of COSMOS IIB was provided and task force committees were formed to cover Field Test, Training, Policies & Procedures, Communica- tions/Video, Software Design & Support, Logistics & Distribution, and Fi nance. Communications Widespread communication within Federal Express became a major ele- ment of COSMOS JIB, particularly between Memphis and the field person- nel. Regular meetings were held by each task force and once a month the leaders- of the task forces would gather to share their progress. Distribution lists on the minutes of meetings grew each month as more and more people became involved. Articles were written for the employee newspaper every month and the monthly employee and manager's videos almost always had a segment de- scribing the operation, progress, and benefits of COSMOS JIB. A special video showing how COSMOS IIB integrated with the many systems at Fed- eral Express was videotaped using a wooden model of the Tracker.
68 CARL NEHLS Any group within the company could obtain a presentation on COSMOS IIB merely by phoning in a request. Almost everyone on the project team made presentations somewhere along the way, but presentations were viewed as such an important function that one individual reporting directly to Dalton undertook them as a major job responsibility. Special attention was given to communications with the field, because the company believes that it is the commitment of the people actually performing the scanning that determines whether or not the project will be a success. The couriers' attitude and willingness to make a project such as COSMOS IIB succeed is often based on their understanding of the value of the project and how they perceive that their local management views the project. To ensure that the interests of the field personnel were properly represented and that the information about the project was properly communicated to field personnel, the ground operations support staff in Memphis was integrated into the development process of COSMOS JIB, and a director of this group was appointed to represent the interests of the field in general. Executive ground operations management, field engineering groups, and local man- agement of test sites were included as much as possible in the ongoing plans for COSMOS JIB, and several trips were made to each test site to provide pretest briefings. The field tests also provided valuable communication opportunities. Videos were made during the tests showing the couriers in action, and the comments of the couriers, both positive and negative, were widely publicized. Testing Production Trackers became available in April 1985 and the first few units were sent to Search Technology in Atlanta, Georgia, for thorough environ- mental testing. Additional units, schematics, and software were sent to Geor- gia Tech Research Institute for an engineering evaluation. Both of these organizations discovered problems with the unit and recommended solutions. It became increasingly obvious at this time that if Federal Express was to be successful in any leading-edge use of technology, it had to become in- volved in the hardware design. Engineers under Turpin developed, with the help of Dr. Mike Maddox of Search Technology, detailed environmental and operational tests. These tests included heat, humidity, vibration, and drops to concrete from as high as 7 feet. Hand Held Products made the required changes and the first units were accepted by Federal Express. Within 2 weeks, in June 1985, couriers were being trained and the field tests had begun. These initial tests looked only at the hardware and software that could be operated within the local station environment; COSMOS was considered a much too important system to jeopardize by connecting it with equipment and software that had never been
CUSTODIAL PACKAGE TRACKING ATFEDERALEXPRE:SS 69 formally tested on the company's quality-assurance systems. Field tests were held in five locations during the next several months and two of those lo- cations continued to use the Trackers even after the tests were complete. A variety of problems were identified and corrected during the 6 months after field tests were begun. Keyboards on the Trackers came loose and the keys stopped functioning, causing great consternation in the couriers. The charging bases had never been tested with more than a few Trackers, and it turned out the power supply was not charging the batteries properly. Con- sequently, the Trackers died on the road and, because the charging bases could not be readily fixed, the test teams had to buy all available supplies of 9-volt batteries from local stores to keep the test going. Because the specified material for the battery terminals would have delayed the test by weeks or months, a substitute material was used which lost its springiness, thus allowing the batteries to break contact occasionally. This intermittent loss of power caused strange things to happen to the Trackers, each of which looked like a separate problem. The field environment proved even tougher than the laboratory, with Trackers being dropped repeatedly, being slammed by doors, and even being run over by a vehicle. Even more important than identifying the hardware problems, the test revealed changes in the software that could make the system dramatically more functional. Couriers were required to answer 12 questions for each delivery exception (such as business closed or package delivered next door), no matter what kind of exception it was. By asking the type of exception and then branching to only the three or four questions required for that type of exception, the software changes could save the courier significant time. Such simple things as making the Tracker ready to scan instead of requiring the courier to turn on the scanner, having the unit always return to the last scan used, and storing information that was often repeated, made the Tracker much easier to use. Many of the suggestions came from the couriers, and after each test a few more changes were incorporated. The tests also revealed some positive aspects about the project. The couriers enjoyed using the Tracker-it made them look "high-tech." Customers wanted to know more about the device, and couriers from competitive pack- age carriers stopped the Federal Express couriers and asked about the "Fedex computer" their customers were talking about. Not only did the couriers enjoy using the Tracker, but the inflation they collected allowed the customer service agents to answer customer inquiries on the initial telephone call. The result was fewer package traces for the station and fewer telephone calls from the couriers reporting on exception packages. Not all of these benefits of COSMOS IIB were free. It took more time for the couriers to run their routes now not a great deal, and newer software was cutting the time as was the learning curve, but the costs could not be ignored. Even before the tests were over the project team was working on
70 CARL NEHLS the financial justification for COSMOS JIB. By this time, however, the environment had begun to change. F· · 7 Nancy Environment By August 1985, ZapMail, Federal Express's 2-hour door-to-door facsim- ile product, had been in operation for about a year and was performing below plan. The operating losses of ZapMail resulted in a shortage of funds for other projects, and so COSMOS IIB was very closely scrutinized. Jim Barks- dale, Chief Operating Officer, requested a complete itemization of all the resources required for full development of the project. The COSMOS IIB team was obliged to take inventory and resell the project to the company. Field management was very concerned about the productivity costs as- sociated with performing both delivery and pickup scans. Performing the delivery scans eliminated some of the couriers' paperwork, but the pickup scans provided no benefit to the couriers for their efforts. In fall 1985 the senior vice president of ground operations agreed that if the Tracker could provide routing information as part of the pickup scan, ground operations would have no objection to implementing all the COSMOS IIB scans. Adding the routing function would, however, require huge additional memory, and that meant a total rework of the Tracker. A distinct advantage of an electronic routing guide to replace the familiar paper guide was the ability to have each origin station route a package differently based on destination. The old method required everyone to route a package the same way. Identical routing worked well under the old hub- and-spoke system of sending all the packages to the Memphis Hub, but volumes in the Northeast Corridor and other areas of the country were large enough that local trucking networks were being implemented, and this prac- tice was expected to grow. The problem with making the trucking network function was recognizing which packages to keep locally and which to send to Memphis. An electronic routing function would provide a code, based on origin location and destination ZIP Code, that would indicate where the package should be sent for sorting. The combined impetus of the desire to make the courier's job easier and the operational efficiencies and flexibilities inherent in the electronic routing feature were seen as enough reason to push ahead for the next version of the Tracker and delay the full implementation. This more powerful scanning device soon become known as the SuperTracker and was marketed by Hand Held Products as its Micro-Wand II. Computing Environment COSMOS IIB required significant amounts of computing power and net- work capabilities. Management saw some clear business synergies to using
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS 71 the already established and currently underutilized ZapMail network for COS- MOS JIB. Consequently, communications and data-processing parts of the project were redesigned based on the ZapMail network. This meant all the programs developed for the MicroVAX station processor had to be rewritten on Tandem computers; consequently, all the applications programmers in Colorado Springs went into intensive Tandem training. The decision to use the Tandem network did not totally eliminate the need for communicating hardware in the station. Some device had to be in the station to charge the SuperTrackers and poll the devices for uploading of data and downloading of software. This device was termed the Smart Base. The actual computing power and control would reside on the ZapMail net- work. Considerable discussion occurred over the virtues of designing a unique piece of hardware for these station functions as opposed to using an off-the- shelf computer, but because of the limited functionality required, the cost advantages, and the growing aversion to distributed systems, the unique hardware solution prevailed. Competitive Environment Another element weighing in the COSMOS IIB decision was the changing competitive environment. By 1985 all of Federal Express's competitors had adopted the hub-and-spoke distribution scheme, service levels were improv- ing, prices were being pushed down, and at least one competitor was doing selective scanning similar to the COSMOS IIA scans. It was becoming difficult to tell the difference between the various services and, thus, more difficult to justify Federal Express's premium price. Besides providing better information to the customer, collecting valuable data on the operations, and making the distribution system more efficient, COSMOS IIB became the lead project in a campaign to use technology as a way to differentiate Federal Express from its competitors. (Two years later, in 1987, COSMOS, DADS, and the SuperTracker starred in a national Federal Express advertising cam- paign.) Implementation Financing The financial plan for the implementation of COSMOS IIB was revised based on the use of the SuperTracker and the use of the ZapMail network. Every department in the company was polled for its requirements. The fi- nancial groups went to great lengths to eliminate duplication of resources and built all the requests into a single package so the total impact of the project could be easily assessed. In January 1986 the entire budget proposal for COSMOS IIB was accepted. This blanket endorsement of the project and unconditional support from
72 CARL NElILS upper management grew from several sources. First and foremost was still Fred Smith's conviction that the information about the package was as im- portant to customer satisfaction as moving the package itself. Now, however, the burden was on the project team. Harry Dalton had been given everything he had requested and there could be no excuses for anything short of a complete and timely implementation. The implementation plan now called for 27,000 SuperTrackers to cover a nightly volume of 500,000 packages. COSMOS IIB would be implemented in two phases. The first phase would begin in October 1986, would require 14 months for complete rollout, and would include only the delivery scans. The pickup scans would be implemented starting in April and would be completed by December 1987. The reasoning for this plan was simple. The SuperTrackers would not be available until September 1986. The delivery functions of COSMOS IIB could be rewritten in that amount of time, but more time would be required to redevelop the other functions. It had also become clear in the field tests that the couriers did better if they were trained on only a limited set of scans. Once they were familiar with the scanner, they could easily learn additional functions without as much effect on their daily productivity. Project Evolution With the scanner in redesign and the applications in rewrite, the project team's attention turned to Smart Base development, a second vendor for the SuperTracker, and a mini-implementation. Hand Held Products had taken all the suggestions and findings from the field studies and built them into a finalized production device it was marketing as a Micro-Wand II with the only difference from the Tracker being the keyboard. Federal Express had already developed a method of collecting the scan data and posting it to COSMOS, and the project management felt a great deal could be learned from having 10 to 20 stations doing COSMOS IIB scans while they waited for the new large-memory SuperTrackers. As a result, 1,100 small-memory Trackers were purchased and in January 1986, Federal Express began its rollout of 19 stations dispersed across the country. The Smart Base began its own development schedule in late 1985. RFPs were distributed and proposals were considered. Norand was finally chosen as the Smart Base vendor. Norand was also chosen as the second source for the SuperTracker. With 19 stations using COSMOS IIB and implementation plans for the rest of the country in full swing, all parts of the company were pulled into the project. Task-force meetings were held weekly, and monthly project reviews with Ron Ponder, senior vice president of the Information Systems Division, of which Dalton was a part, had to be held in auditorium-size
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS 73 conference rooms to handle the number of people attending. Dalton was regularly required to provide formal briefings to senior management and occasionally to the board of directors. Dietzel's staff was soon doubled to 14 to help monitor the stations already on-line and to manage the continuing development effort. The monthly review meetings with senior vice president Ron Ponder typ- ically lasted a half day and included reports from all areas of the company. Although the meetings finally became so large that they had to be broken down into submeetings, they served a vital purpose. Federal Express had adopted the practice of developing a separate computer system for each new project, such as billing, payroll reporting, and field reporting. These systems had developed different ways of identifying locations and in some cases had even begun to use the same piece of information in different ways because the exact meaning of the information was not clear. COSMOS IIB was designed to deliver information to all of these systems and each system had to be made consistent with the others. Most of these systems were in the Information Systems Division, which made it easier to see the discrepancies and make sure that they were reconciled. Ron Ponder was also compelled on several occasions to rally the troops and keep the project moving. As with any project, stumbling blocks were hit, and each group would press for a little more time to make its adjustments before moving on. The pressure to meet the implementation date, however, dictated that the process be kept moving. As problems arose, action plans were created and responsible people were appointed to make sure that the plans were implemented on time. COSMOS IIB became a top priority throughout the Information Systems Division, and Ponder was quick to supply whatever resources were required to keep COSMOS IIB on track. A New Direction By mid-1986 it had become evident to Federal Express that problems with the ZapMail network were causing service commitment problems. Many of the communications links were land-based telephone lines that were difficult to get installed on time and were often not available in the capacity required to support the ZapMail data flows. Not being able to provide top quality service defeated the purpose of the product, and the Challenger disaster put the ultimate solution of flying a Federal Express satellite on indefinite hold. Top management decided that it was not in the best interest of the corporation to further burden the network with COSMOS IIB data. In August 1986 Ron Ponder called a special all-day meeting to review the requirements of changing from the distributed, Tandem-based ZapMail net- work to a centralized IBM solution using the company's IBM network. All groups were to give a detailed synopsis of the impact to ensure that the rollout
74 CARL NEHLS would continue as planned. Once it was decided to move away from the ZapMail network, the decision to use the IBM solution was fairly easy to make. One of Federal Express's greatest strengths was the large base of IBM expertise it had developed over the years. Many of the problems COSMOS IIB might experience had been solved before. Probably the largest impact of the move to the IBM solution was the requirement to again rewrite the software for the COSMOS IIB reports. The software development for the Tandem computers was nearly complete. All this work was scrapped and the application programmers were sent for train- ing on IBM's operating system. The Smart Base also experienced significant modification because of the decision to move to the IBM/SNA system. One of the primary functions of the Smart Base was to communicate on the network, and the protocol it used for this had to change entirely. Also, the local computing power to actually control the SuperTrackers had been eliminated with the move away from ZapMail and the centralized mainframes could not provide this function. What had started as a power supply with a simple 10-key pad and 2-line display used for hardware polling was turning into a full computer complete with floppy disk drive, 750K of memory, and applications software for loading local parameters (see Figure 71. At this same time the SuperTracker programmers in Colorado Springs raised an alarm that the ever-increasing demands on the software were stretch- ing the capacity of the SuperTracker EPROM. The changes required to make the SuperTracker easy to use and very functional had grown to over 30K and more changes were seen on the horizon. It was recommended that the software be written to operate in RAM instead of EPROM. This afforded almost limitless room for applications software, but there was a risk that static charges, voltage peaks, or magnetic fields could cause the software in the SuperTracker to disappear. If the software was lost, the SuperTracker would become useless until it could be returned to the Smart Base and have the software reloaded. The benefits of unlimited space for software were considered to vastly outweigh the unlikely occurrence of software loss. An- other big advantage to RAM-based software was that the SuperTracker soft- ware could be downloaded from the mainframe computers in Memphis to the Smart Bases in the stations, and from there into the SuperTrackers all automatically. This feature would make changes to the software very easy to make as none of the SuperTrackers had to be specially handled. Problems and a Delay There were continuous problems and changes, which could not all be solved by action plans and resources. With great reluctance, the project team petitioned for a one-time delay to the implementation schedule. A new im
CUSTODIAL PACKAGE TUCKING AT FEDE~E EXPRESS 75 FIGURE 7 Norand Smart Base. plementation starting date of January 1987 was agreed to, but the imple- mentation completion dates of December 1987 for Phase I and May 1988 for Phase II were frozen. These completion dates soon became the target. If something needed more time to be developed, then something else had to give up its time so the completion dates could be met. A variety of problems were hardware related. For instance, the Super- Tracker hardware design was often ahead of the chip manufacturers' devel- opment schedules. As an example, the SuperTracker manufacturer was waiting for the chip manufacturer to certify a surface mount version of their low- power consumption static RAM for vapor-phase soldering so production could begin. In general, most hardware delays put a strain on internal re- sources at Federal Express as the engineers worked as hard as vendors and suppliers to overcome the problems. Other hardware problems also arose with the SuperTracker clone. After a promising start with several innovative ideas, the development effort was soon mired down in difficulties meeting Federal Express's specifications.
76 CARL NElILS The hardware configuration for the clone was essentially the same as in a piece of existing equipment, and the changes required were a bit too excessive to make the unit operate up to snuff. Federal Express was finally compelled to drop the development of the SuperTracker clone. The changing business conditions had compelled the project team to request frequent changes in the requirements of Smart Base. Norand was having problems fitting these changes into the original design, and the modifications were causing significant delays to the Smart Base implementation date. These delays caused a group in Colorado Springs to initiate a separate effort to develop the Smart Base functions on a personal computer. SuperTracker Implementation Finally, in January 1987, 3~/: years after the issuance of the REP, contracts had been modified and approvals obtained to purchase and implement 30,000 SuperTrackers to cover nightly volumes of 750,000 packages. Training for the couriers and ground operations managers became a major task. The couriers had been hearing for 2 years that COSMOS IIB was coming, what it looked like, and how it would work. But, teaching 20,000 people how to operate a device even one with software so refined and structured that it would not allow them to make a mistake-was a monumental task. The final plan was to train professional trainers who would travel around the country training operations management and volunteers from stations called information coordinators. These information coordinators would then return to the station and teach the other couriers. The information coordinators were given a full week of training including practice on teaching skills. When they returned to the station they gave 4-hour classes to groups of from 12 to 15 couriers at a time. Although the rollout of COSMOS IIB had begun, the development was far from over. Before the SuperTrackers could be installed in non-DADS locations, the Smart Base had to work so the couriers could transmit their scans. Smart Bases with charging capability, but no communications ability, were developed. However, fully functional units for non-DADS locations appeared to be a long way off. Acoustic transmission by the couriers was seen as an alternative way to implement non-DADS locations, but the mo- dems were still not operating properly. Soon the rollout plan was modified to put all the non-DADS stations at the end of the implementation. The SuperTrackers were implemented with delivery scan software residing in EPROM for several reasons: (1) it had already been written, (2) the RAM software had not been completed, and (3) the Smart Bases required to load the RAM software into the SuperTrackers were not available. Lack of a working Smart Base would have prevented RAM-based software from being implemented in January, but the fact was that the software had
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS 77 been caught up in some changes of its own. Shortly before the first version of RAM-based SuperTracker software was to be released, the risks associated with total loss of software were reassessed. It was felt that having a totally useless SuperTracker was unacceptable; a backup version of the scans had to be kept in EPROM so that even if the RAM software were lost, the courier could continue to carry out the basic elements of the scanning process. Several meetings ensued and a pared-down version of the scans was agreed to as was a new schedule. A major rewrite of the SuperTracker software was required to accommodate the new plan. A Smart Base Breakthrough The Smart Base looked as though it would become the nemesis of COSMOS JIB. Fitting the broader requirements into what had begun as a piece of equipment with limited functions seemed nearly impossible. But, a working Smart Base had to be available for the project to succeed. The word was out that the group working on the PC-based "bootleg" version of the Smart Base had achieved a very functional unit using all off- the-shelf hardware and some already developed network protocol software. This "PC Base," as it came to be called, had not been tested on the quality assurance system, but it seemed to hold a lot of promise. Work on the PC Base intensified to ensure that a solution would be available to support the implementation schedule. In July 1987, as the PC Base was completing its quality assurance process, Norand successfully delivered the working Smart Bases. Federal Express now had two working solutions, but Norand would require several months to supply its bases in quantity and the unit would still have to go through Federal Express's quality assurance process. Consequently, Federal Express decided to use both solutions. The PC Base used the same computer that the Customer Automation Group was purchasing to use as customer meters. (A customer meter helps the customer automate his or her shipping process by generating airbills, gen- erating reports, and providing paperless billing information to Federal Ex- press.) A deal was made with the Customer Automation Group to "borrow" several hundred of their computers to use as PC Bases. When the Norand Bases became available, the computers would be returned to the Customer Automation Group. Future Table 1 presents a summary of the development of COSMOS IIB from 1977 through 1987. In early 1988 COSMOS IIB is still rolling out, is still evolving, and is still being developed. PC Bases are in the stations and some
78 TABLE 1 COSMOS IIB Timeline: 1977-1983 CARL NEW 1977 COSMOS implemented. 1978 Bar code is investigated as a way to collect package tracking numbers. 1979 COSMOS rewritten for IBM equipment. Dalton moves to Colorado Springs and begins to test bar-code scanners. Concept of terminals in vans for dispatc~-;rng is conceived. 1980 COSMOS II is split into COSMOS IIA and COSMOS JIB. 1981 COSMOS IIA is implemented. 1982 First DADS systems implemented. 1983 Proposals for COSMOS IIB scanner are solicited. 1984 Funds are approved to develop and test COSMOS JIB. 1985 Trackers are field tested. Decision to use ZapMail network. Smart Base planning initiated. First pass at computerizing customer and dispatch information. Concept of recording package numbers at each change of possession first . . envlslonec V. Larger mainframe capacity required. First test of concept shows information can be collected, but better scanners are needed. Dispatch volumes growing and logistics of voice becoming more difficult. COSMOS IIA only scans packages in the station; IIB will scan packages on road with tougher hardware. Norand equipment is selected and then modified to meet Federal Express's needs. Packages are scanned as they arrive at the station and as they leave the station. Scans are sent to COSMOS to provide customer information. Radio channels too congested for voice dispatching; digital messages are sent to terminals in the vans. Near real-time link of Federal Express computers to courier van is established, the more recent the information the higher the value. Established companies are not listening to Federal Express's needs. Hand Held Products agrees to build exactly what is requested. IIB scanning device called the Tracker. Specifications evolve from a simple scanner, to a scanner with an LCD display, to a unit with a display and a simple keyboard to a final design with a 2-line display and a full keyboard. Many hardware problems discovered and solved. Software changes make the unit much easier to use. Station computer and developed software are dropped. Using the established ZapMail system provides business synergies. With no station computer, a separate charging unit with hardware controlling functions had to be developed.
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS TABLE 1 Continued 1985 Decision to wait for large memory Trackers. Plans to develop a second source for Trackers are made. 1986 1,100 Trackers are implemented as a major test. Decision to stop ZapMail. Development of the SuperTracker II is discontinued. 1987 SuperTrackers are implemented for delivery scans. PC Bases are installed and Norand provides working Smart Bases. 79 More memory allows electronic routing that enables a more efficient distribution system. Federal Express's executive management was not comfortable with the success of the project so heavily dependent on a single vendor. Using small memory Trackers allowed the system to be exercised and provide data to develop a beuer ultimate system. Without the ZapMail network the programs had to be rewritten to run on IBM. A second source of SuperTrackers could not meet the specifications. 25,000 SuperTrackers are implemented nationwide to scan packages as they are delivered. PC Bases, started as a bootleg project, are used to keep the rollout moving while Norand completes last-minute changes to their production unit. have already been replaced with Norand Bases. Equipment challenges still exist with the acoustic transmission not working as reliably as desired and with future plans to have customer meters and SuperTrackers exchange in- formation on a daily basis. Special scan types are being developed for Federal Express sort locations and international-customs operations. These special scans and the accompanying reports will soon bring the benefits of COSMOS IIB to every area of the company. COSMOS IIB is actually a window to the future. Even before the project is fully implemented, new products are being developed that would not be practical without the information provided by COSMOS JIB. As the field personnel become experienced with COSMOS JIB, they are quick to identify new applications to help them in their day-to-day operations. Ideas never conceived when COSMOS IIB was designed will continue to enhance the value of the project for many years. CONCLUSION COSMOS IIB was driven by the concept that the information about the package is as important as the package itself. In fact, the objective of pro- viding real-time tracking and tracing information about the package even
80 CARL NEHLS tually became a requirement for long-term viability in the marketplace. Because the vendors of bar-code scanning equipment would not provide the proper equipment required to capture the information, Federal Express took it on itself to have the scanners developed. Federal Express soon discovered that developing specialized hardware with a vendor was a skill in itself. It took 3~/: years of working with a vendor to finally define, develop, and take delivery of a scanner that met all the needs of the corporation. However, after acquiring the internal expertise for de- fining, testing, and monitoring development, a similar hardware project, in the form of the Smart Base, took only about a year to complete. Even when the equipment was defined and being manufactured, it was a major challenge to the project team to keep the process moving and on target. Continually changing circumstances required multiple iterations in virtually every aspect of the project, particularly in the writing of software. Optimism frequently gave way to frustration as the rules changed or were reinterpreted. The ability to react, adapt, and keep on moving became the hallmark of the project team. COSMOS IIA experience provided important guidelines in the confusing maze of a constantly changing COSMOS IIB program. Complex decisions about what to implement and how quickly to implement were always arising. COSMOS IIA was a constant reminder that pieces of the system could be added, but the sum of the pieces had to represent a self-sufficient system. For example, delivery scans with only DADS to communicate the information could be implemented, but those scans had to have management reports before they could be integrated into the corporate package tracking and tracing system. Implementing pieces of the project was considered part of the "don't let best get in the way of better" philosophy of Federal Express. These pieces often required some reworking as later parts were added, but this was not necessarily negative. The experience gained from each new piece helped direct the future phases of the project to an overall better solution. Even entire report programs that were never implemented because of a change in the system hardware design provided new ideas for better data-base designs and report layouts. This is not to say that the project team preferred to move in bits-and-pieces instead of implementing a total project. However, given the scope of the project and an environment that could not be controlled, there was no choice but to keep what was good and implement what could be developed in the time available. Despite its trials and tribulations, COSMOS IIB is considered one of the best managed technical projects ever developed at a company that has always pushed to use technology. Federal Express has adopted a learn-as-you-go methodology that allows the company to benefit from the mistakes as well as the advances that it contributes to the technology it is attempting to use.
CUSTODIAL PACKAGE TRACKING AT FEDERAL EXPRESS 81 Harry Dalton sums up this general approach in his oft-heard comment- "The only thing we know for sure is that we won't get it quite right the first time, but we'll come back and fix it." ACKNOWLEDGMENTS I would like to express my appreciation to David Dietzel, Jim Turpin, Pat Mulvey, Chris Demos, Arun Kumtha, and Chuck Theobald for their direc- tion, input, and support in the preparation of this chapter. My special thanks to Harry Dalton for sharing his project management insights, his unwavering intent to provide a factual, instructional case history, and his overall partic- ipation in making this endeavor a success.