In recent years, annual supply chain losses in the retail industry have ballooned to
$70 billion. Amazon.com’s supply chain losses account for $250 million of that. While
$250 million may seem small compared to $70 billion, it still warrants attention because
supply chain losses account for nearly 10% of Amazon.com’s annual costs. Recent
studies conducted have shown that most of the losses incurred in the supply chain result
from a poor product tracking system. Outraged by the high losses, Amazon.com’s CEO
Jeff Bezos was prompted to set aside $1,000,000 for his Supply Chain Management
Group (SCMG) to evaluate Amazon.com’s barcode-based product tracking system. The
specific goals set forth for this study were to establish a standard set of criteria for
evaluating the current and all future product tracking systems used by Amazon.com and
to recommend any changes necessary in order to minimize supply chain loss.
This paper presents the findings of the study. First, other product tracking
systems were studied, and three criteria – security, scalability, and ease of use – were
chosen to be evaluators of Amazon.com’s own tracking system. However after
evaluation, it was found that Amazon.com’s barcode system did not satisfy the security
and scalability criteria. Thus, the SCMG went forward and researched a replacement
system that would satisfy all the criteria. The solution selected by the SCMG was Radio
Frequency Identification (RFID). RFID tags are small microchips that can store and
wirelessly transmit information. We in the SCMG recommend that Amazon.com replace
all barcodes with RFID tags because RFID is secure, scalable, and easy to use. This
paper serves to show that the barcode tracking system is in fact flawed, and that an RFID
tracking system should replace it.
2.0 Criteria for Effective Product Tracking
The SCMG studied product tracking systems used by other leading retailers to determine
a set of criteria for evaluating tracking systems. After months of study, it was found that
the retailers with the lowest losses had product tracking systems with security, scalability,
and ease of use, while the companies with the highest losses lacked all three.
A perfectly secure product tracking system would be able to ensure that what a
manufacturer thinks is going through the supply chain is what is actually in the supply
chain. It would eliminate the possibility of theft or misplacement of any product in the
supply chain. Techniques used by the studied retailers included automatic notification
the instant a loss occurred, detection of fraudulent product identifications, and prevention
of notifying a supervisor whenever a product has been misplaced. This allows the
supervisor to pinpoint when and where a loss has occurred so that he/she can correct the
problem immediately, preventing the same loss from occurring in the future.
A scalable product tracking system can adapt seamlessly and inexpensively to expansions
in inventory. Scalability is usually determined by finding the additional resources needed
per unit of expansion. That is, how much it will cost a company per product to expand its
inventory. Obviously, the lower this figure is the better. While a grocery store with
stable inventory sizes may not be interested in this criterion, a company such as
Amazon.com with a volatile inventory must consider scalability.
Take for example Amazon.com’s new jewelry and sport shops. The introduction of these
shops this past month added hundreds of thousands of new products into Amazon.com’s
inventory. Every single one of these products has to be identified and added into the
tracking system. If the tracking system in place was not scalable, then the costs
associated with integrating each product would be too high in order for Amazon.com to
make a profit from the expansion. However, a high scalable system keeps the costs per
expansion low, almost making the expansion transparent to the eyes of the head
2.3 Ease of Use
A system must be easy to use in order to minimize costs. It costs money to take the time
to train a worker to use a system. If the system is easy to use, then little time will be
needed for training, and thus less money will need to be spent. In addition, mistakes also
cost money. Whenever a worker makes a mistake, time and money must be spent in
order to correct it. Not only does the mistake need to be fixed, but the worker who made
the mistake will need to be retrained so that the same mistake is not made again. All this
can be avoided with a system that is easy to use. The probability that a mistake is made
lessens as ease of use increases.
3.0 Pitfalls of Barcode
Given the three described criteria, the SCMG conducted a study of
Amazon.com’s current barcode-based product tracking system. The results of the study
showed that the barcode system does not meet all of the criteria. The barcodes in use by
Amazon.com consist of lines that represent a small string of numbers usually between 5
and 20) which are printed on paper. The lines can be read by an optical scanner, which
then relays the string of numbers to a computer to retrieve pertinent information. While
this results in an easy-to-use system, by no means is it secure or scalable.
In a barcode-based system, a product’s identity is solely dependent on the printed
lines. Since these lines are simply printed on paper, this leads to easy counterfeiting.
One barcode can be easily photocopied, hand-edited, or damaged in order to match
another. In a tracking system that solely relies on scanning barcodes, these counterfeits
can lead to security problems. This is because barcode-based systems in actuality do not
track products, but instead track barcodes. Only the barcode of a product needs to be
present in order for the system to think that the entire product is there. So because
barcodes are easily counterfeited, a barcode-based tracking system can be easily fooled
into thinking that the counterfeited barcodes are actual products.
Not only do barcodes lack security, but they are also not highly scalable. Many
may argue that because barcodes are easy to print, they are also cheap. This is true.
However, the scalability issues from barcodes arise due to the limited information that
can be stored on a barcode. As stated above, barcodes typically contain a string up to 20
digits and take up a 1” x 1.5” footprint to do so. These numbers are great for representing
a model number, manufacturer identification, or product number. However, it is
impossible to expand barcodes to identify individual items by their serial numbers. Serial
numbers alone can contain up to 50 digits. So in order to be able to track individual
items, barcodes would have to contain 70 digits, resulting in a footprint of 1” x 5”. Many
product packages are less than 5” across. So not only would the barcode size have to be
expanded, but product packaging would also have to increase. This is simply too much
trouble for what it is worth, which is why up to this day barcodes still do not contain
serial number information.
4.0 Radio Frequency Identification as a Replacement to Barcode
Due to the inadequacies of the barcode system, further studies were conducted by the
SCMG to find a replacement system that does satisfy all of the criteria. After speaking
with Walmart CEO Jeanne Jackson, it was determined that RFID is the best solution.
4.1 What is Radio Frequency Identification?
RFIDs are small microchips that can store securely and transmit wirelessly digital data.
The microchips measure 0.4mm x 0.4mm, or 1/100 of a fingernail. The microchip is
attached to a spiral-shaped antenna that functions as a radio signal receiver and
transmitter. Including the antenna, the entire RFID device measures 3cm x 3cm.
4.1.1 Technical Macroscopic Overview
An RFID takes advantage of technologies developed in the computer processor industry
for its microchip. Essentially the microchip on an RFID is a super miniaturized computer
processor. Half of the chip is reserved for data storage. An RFID can store up to
2kBytes of binary data (214 1’s and 0’s) using these transistors. This translates to over
1000 digits of storable data. This data is written into the RFID microchip at
manufacturing and cannot be altered afterwards without destroying the chip. About 10%
of the chip is used for power. This power unit converts radio signal energy into electrical
energy. Even though radio signals are generally low in energy, it is still enough to power
an RFID because a typical RFID operates on only 0.3mW of power. This is 50000 times
less power than an Energy Star compliant 15W fluorescent light bulb. Another 30% is
used for signal decoding. A series of logic gates are used to decode the signal received
by the RFID into a form understandable by the RFID. As with the data, these logic gates
are embedded into the microchip when the RFID is manufactured and cannot be altered
afterwards without destroying the RFID. This also serves as a security check because
only genuine RFID’s will be able to properly decode the signal.
An RFID tag is a miniaturized, self-powering, wireless memory module. It stores
data that can be accessed by a wireless sensor up to 30’ away. (6:114) It is implemented
with a 0.4mm x 0.4mm microchip and connected to a radial antenna. The chip is divided
into two circuits: digital and analog. The digital circuit serves as the brain of the RFID,
storing, decoding, and encoding data. The analog circuit serves as the brawn, providing
power and modulating signals. (7:46-47) Figure 2 shows a sample schematic of an
Figure 2. A sample schematic layout of an RFID chip
(Source: Takaragi, K., et al. “An Ultra Small Individual Recognition Security Chip”, IEEE-
Micro, Vol. 21, No. 6, Nov/Dec 2001, p. 47)
2.1.1 Digital Circuit
By using 0.18µm transistor technology, the digital circuit provides a compact
solution for data storage. Data is stored as bits in read-only memory (ROM), with each
bit corresponding to one transistor – “on” for 1, “off” for 0. Typically an RFID stores
64-256 bits of data. Because the feature size of the transistors is only 0.18µm, the
amount of storable data can be easily customized without significant change to the size of
the RFID. The data is written into the ROM when the RFID is manufactured, and cannot
be modified afterwards. (7:47)
The decoder ensures that the RFID can only be accessed by corresponding
sensors. It is composed of a series of complementary metal oxide silicon field effect
transistor (CMOSFET) logic gates. When a signal is received by the RFID, the signal is
passed through the logic gates to be tested for authenticity. The logic gate network, like
ROM, is programmed into the RFID during manufacturing and cannot be modified.
(7:46-47) This thwarts all unauthorized used of the RFID. If the signal does not pass the
tests, then it was not authentic, and the RFID takes no action. If the signal is determined
to be authentic, then the data stored in ROM is encoded by being fed backwards through
the logic gates and sent to the analog circuit for transmission.
2.1.2 Analog Circuit
The analog circuit takes advantage of the energy storing and filtering capabilities
of a capacitor to power the RFID. (7:47) Figure 2 shows a model of the circuit. The
antenna, which functions as an inductor, creates current when it picks up a radio signal
from a sensor. This current is used to charge the capacitor, which then discharges power
into the digital circuit. When data leaves the digital circuit as current, it is passed back
through the inductor, which generates a radio signal that is transmitted back to the sensor.