In a retail
test, a new generation of computer based training
significantly reduced produce shrink at checkout.
This is the first of two columns describing the
experiment.
|
|
Customized
Learning
(©Supermarket
Business, September 1997)
by: Richard
Shulman
I've recently had the
exciting experience of seeing the future of computer
based training (CBT) and the enormous potential it can
offer your company. This month and next I plan to share with you the
methodology used, the experience of the retailer where
the initial research was performed and the technology
that was used.
|
|
First, the
method and the technology were developed by a special
group at NCR called the Human Interface Technology
Center. This organization has the corporate
mission of making computers and computer systems easy to
understand, easy to use and effective in accomplishing
tasks. The goal of this group is to create
computer-based systems so intuitive that intended end
users can rapidly learn to operate them by just studying
the features of the interface.
|
|
NCR works
with companies they refer to as "Lighthouse
Customers,: innovative industry leaders. In the
food industry they have worked with some premier retail
organizations.
|
|
In the
current project, code named Advanced Produce Interactive
Cashier Off-line Training System, or
"Apricots," NCR is working with Balls Food
Stores in Kansas City. Balls operates more than 20
outstanding supermarkets under the Balls, Hen House and
Price Chopper names. The technology uses an
approach referred to at NCR as "see it, hear it, say
it" to imprint information on the student.
From a technology perspective this is the first use I've
seen of voice response technology in computer based
training.
|
|
In the
Apricots project the NCR team had a number of
goals. They needed to develop methodologies to
estimate produce shrink due to identification
errors. They wanted to explore computer based
training techniques and to use leading edge technology to
customize a learning system for produce. Finally,
they wanted to test this system and determine the
business rationale for deploying product versions of the
technology.
|
|
The test
design that was used to design and evaluate the Apricot
system at Balls is so significant that I think it should
be carefully evaluated by every retailer seriously
considering getting involved in CBT. The purpose of
the test design is to allow the retailer to evaluate his
current level of performance and then evaluate the impact
of using the new technology. In this situation
Apricots was designed to improve the accuracy of produce
recognition.
|
|
The first
step in the development was to completely understand the
current process by which a cashier at Balls determines
the PLU for a product. Balls has a large produce
department, typically carrying more than 225
products. Its display technique is predominantly
bulk, and only a small percentage of the products are
scanned, i.e. bagged salads, etc.
|
|
Balls has
implemented the PMA standard coding, and as one would
expect many products arrive at the store
source-PLU-marked. Balls has a policy that the
cashier must read the PLU sticker through the plastic
bags and not directly handle the customer's
produce. After careful observations and interviews,
it was determined that the cashier typically attempts to
recognize the product and enter the PLU without looking
for it on the product.
|
|
The
explanation frequently given is that many products--
peppers, for example-- don't retain the stickers and that
sometimes the cashier recognizes that the customer may
have switched stickers.
|
|
In the
extensive Balls produce line many products, such as
exotics, leafy and root vegetables, may not carry a PLU
sticker. If the cashier enters what she believes
to be the correct PLU and it's a valid number (accepted
by the POS system), the transaction is considered
complete, albeit possibly incorrect.
|
|
If the
cashier doesn't recognize the item, she has a reference
chart to assist her. If she thinks she knows the
name of the product she uses the chart, and that decision
may or may not be correct, but if the PLU is valid the
transaction is complete. As a last alternative, if
the cashier has no idea what the product is she asks the
cashiers nearby and accepts the consensus identification.
|
|
The bottom
line is that while the adoption of the PMA standard is a
major step forward in the management of a produce
department, it does not assure any retailer of optimizing
his potential gross margins. The Apricots project
was designed to measure how training with advanced
technology could address that opportunity.
|
|
This
project's first strategy component was to focus on the
most important produce items, with most important meaning
those that contributed the highest gross margin dollars
to the department. At Balls the project team
collected a year's purchase history, by item, with the
costs and retails charged. Using this data they
were able to develop a list of the top 148 non-scannable
items, ranked by gross profit dollars. Ten of those
items were projected to contribute more than three
percent each to the department's total gross profit.
|
|
Among the
items on the list were green bell peppers, bananas,
cucumbers, broccoli, peaches, green onions, red seedless
grapes, romaine lettuce, grapefruit and cantaloupes.
|
|
To establish
the current level of cashier knowledge about produce, a
careful study was made of all 500 cashiers. The
analysis was to determine the composition of the
population, e.g., the seniority and level of produce
knowledge.
|
|
This
information, representing almost 12,000 answers,
represented the base case for the training process.
It indicated that the cashiers were correct on either
name or PLU more than 45 percent of the time and that the
accuracy of novices with more than 16 weeks of on-the-job
experience was comparable to that of the experts.
The question was how to quantify the 37 percent of the
"I have no idea" answers.
|
|
By reviewing
the process the cashier would go through and modeling the
responses given by all the cashiers in the test, the NCR
team was able to assign an error rate to these questions
and an error value in dollars. When all errors were
quantified and translated into dollars, the current error
rate represented a lost profit opportunity that was
approximately nine percent of sales!
|
|
The second
phase of the Apricots project was to train a sample of
cashiers and evaluate the results. Another sample
was chosen comprising novices, control novices and
control experts. Each group was given a similar
test, and the base case for the training phase was
established as shown in the accompanying table:
|
|
BASE
CASE FOR THE TRAINING PHASE
|
|
*
Total of overcharges and undercharges to customers
|
|
Using the
Apricots training environment, which I will discuss at
great length in next month's column,
each cashier in the novice group used the computer based
training system developed by NCR for an average of 12
sessions of 40 minutes each. When each novice had
successfully completed the training program, they were
re-tested using pairs of high quality photographs of the
produce, except in this test the novices were tested with
and without the checkstand reference chart.
|
|
The results
were striking. They show that the error rate for
the novices, when using the checkstand reference chart
coupled with the Apricots training, was reduced from 9.8
percent of sales to 1.8 percent. If all cashiers
working at Balls had been trained, the results could
translate to an improvement in gross margin dollars of
more than five percent. If the cost of the
associated training time was measured against this
improvement, the return on investment would be less than
two weeks. The use of the Apricots technology
represents an incredible opportunity for any retailer.
|
|
While shrink
reduction is the focus of financial measurements, the
more appropriate measurement is total errors, the sum of
mistakes that result in both overcharges and undercharges
to the customer. If your training priority is
focused on shrink reduction, you should determine your
company's critical products by following the NCR
methodology to identify your (gross profit) critical
products. I don't have the space in this column to
present the step-by-step process of developing the
measurements used by the NCR team. I would hope
that in the future some trade organization will publish
this in detail as an industry guide.
|
|
At this
juncture it is important for the reader to understand
that there are a number of companies already offering
produce recognition systems that use CBT. They
include Payback Training Systems, Interactive Supermarket
Technologies and Strategic Systems. The effective
implementation of training tools from any of these
companies should result in a reduction is cashier
recognition errors and an improvement in gross profit.
|
|
The
technologies and approach used by the Apricots prototype
are unique, and next month I'll
share the approach with you. However, regardless
of the technology you select, your company must be
committed to training-- a commitment demonstrated by
management execution, not management policy.
Training must be a company "religion" and store
management must reinforce and spend the training time
required for every employee.
|
|
Next month I'll show you the
components of the Apricots learning process and share
with you how voice recognition technology will change the
way we train our associates and operate our supermarkets.
|
