|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 36
Pictures and Parts:
Delivering an Automated
Automotive Parts Catalog
FREDERICK A. FELLOWES AND DONALD N. FREY
This case study reviews the experiences of Bell & Howell's Publication
Systems Division (PSD) from the late 1970s until the present in successfully
developing an automated parts catalog lookup system. * It examines the trans-
formation of a microfiche publishing service business in applying new and
unique electronic technology to meet its customers' information retrieval
requirements.
The case also illustrates the critical success factors of continued and intense
focus on customer needs; application of new technology to meet customer
needs; persistence in spite of failures of several products; the necessity of
active, ongoing corporate support; and, finally, strong project and program
management of new technology for PSD.
EARLY HISTORY
One of the world's largest micropublishers, PSD converts thousands of
catalogs, manuals, and price lists to microfiche. Organizations responsible
for high-volume communications with hundreds of dealerships, branches,
and field-support facilities often rely on micropublishing to reduce com-
munications to a manageable size and to controllable costs. By dramatically
reducing printing and mailing costs, micropublishing enables large documents
to be economically distributed frequently, providing dealers and service op-
erations with more current information. Producing more than 10 million
* Prior to March 11, 1985, the name of the Publication Systems Division was the Micro
Photo Division. For purposes of this case study the Division will be referred to as PSD.
36
OCR for page 37
PICTURES AlID PARTS
37
microfiche a month-the equivalent of 2.5 billion pages of printed infor-
mation PSD's business mission is to continue as the market share and
quality leader of technical document republishing.
Parts Catalogs and Service Manuals
Most manufacturers of vehicles and heavy equipment use a mix of paper
and microform documentation to support their own or their dealers' repair
and parts businesses. Documentation is essential to parts sales and customer
satisfaction. Parts sales top $1 billion annually for several automakers, and
for many farm equipment dealers, parts and service revenues exceed new
product sales.
One set of parts catalogs for General Motors cars requires 16 feet of shelf
space and larger dealers may subscribe to 10 or more sets. Each set consists
of more than 20,000 pages of illustrations, 15,000 pages of charts, and nearly
one-half gigabyte of ASCII text (filling additional tens of thousands of pages) .
Larger data bases are common the parts catalogs for Caterpillar contain
nearly five times as much documentation.
New editions of a parts catalog are published, on average, every four
months and mailed to the dealership. In paper form catalog distribution
(mailing) costs are substantial. Between editions, corrections and other changes
are sent as bulletins, and the number of bulletins can be very large.
Manufacturers choose to publish their parts catalogs on microfiche rather
than paper mainly to save costs. End-users' attitudes range from neutrality
to dislike of microfiche as compared with paper (owing to legibility and
convenience factors).
The Creation of Parts Catalogs
The process of creating new parts catalog editions evolved simultaneously
with data processing. By the early 1970s computerized photocomposition
for parts catalogs was popular. Changes to the text were identified and entered
for each individual catalog; then photocomposition tapes were written which,
in turn, were used to print the catalog. In this form, microfiche versions of
the catalog could be produced directly, but the relationships between data
elements necessary for an "electronic parts catalog" did not exist.
As data processing techniques for relational data base management spread
to publishing in the mid and late 1970s, a parts catalog data base became
an important way of lessening the time and labor content required to update
a set of catalogs, as well as a way of improving accuracy and consistency.
The basic data base element is the parts record containing all applications
of the part across all catalogs. A part change or addition, although affecting
many catalogs, had to be entered only once. Essential to a commercially
OCR for page 38
38
FREDERICK A. FELLOWES kD DONALD N. FREY
viable electronic parts catalog, the existence of a parts catalog data base is
referred to as an "automated" catalog.
The processes of creating a publication edition are known as "prepress"
activities because they are the steps taken before the printing press makes
copies or the micropublisher (such as PSD) generates copies on microform.
Micropublishing Processing
The data processing for parts catalogs is very different from typical man-
agement information system operations. Data bases typically-contain 100 to
500 megabytes of ASCII data. The data base also includes thousands or tens
of thousands of images a data type foreign to nearly all data processing
operations. Even the control and format encoding of text is different-"pho-
tocomposition" files have their own unique rules.
Text data are received on reels of standard ~/~-inch, 9-track magnetic tape.
The text data are reformatted, resequenced, and indexed. Art is received in
three forms: paper, raster data on magnetic tape, and vector data on magnetic
tape. Paper copies are photographed, and image data records from magnetic
tape are scaled and sequenced. The art and text are then combined and fiche
masters are produced. Masters are then used to produce duplicate fiche for
distribution. Silver halide-based film duplicating is capital intensive, whereas
diazo and vesicular film duplicating are not.
In the 1970s PSD was regarded as the "Cadillac" of micropublishing
service bureaus in terms of product quality, service, and indexing. PSD was
one of the few service bureaus to use all three microfilm technologies (ve-
sicular, diazo, and silver) to meet customers' technical documentation du-
plicating needs. Recognized as the best technical document micropublisher,
PSD had acquired a substantial market share. Its customers were primarily
automobile, construction, agricultural, and computer equipment manufac-
turers.
The 1632 Project GM I
In the mid-1970s, through the efforts of Bell & Howell's Document Man-
agement Products Division (DMPD), PSD became aware of GM's desire to
improve parts catalog lockups. GM was interested in improving its auto-
mobile dealerships' ability to retrieve parts catalog information.
John Marken, the division president at the time, recalls:
The general sales manager at GM wanted a fast automatic frame selection microfilm
system. When we learned about the situation from DMPD, a competitor was far
ahead of us. They had already proposed a cartridge microfilm system. Dick Miller,*
* Dick Miller, past president of DMPD, started Bell & Howell's parts catalog micropub-
lishing business (when it was part of DMPD) in the 1960s and was well known to GM.
OCR for page 39
PICTURES AND PARTS
39
from DMPD, did some superb selling to slow down the process and provide us with
time to develop a concept. In 1976 the GM 1632 project was assigned to PSD from
DMPD. We proposed to GM management that we would introduce more automation.
We would use a "roll" of microfiche, for the entire GM parts catalog, giving them
more information, and a way to get to the image faster than with a multitude of
microfilm cartndges. We convinced them of the concept. We built 125 units and
did a field test. The result was favorable.
In 1978 the 1632 rollfiche reader concept was presented to GM's Dealer
Council in Brownsville, Texas. John Marken remembers dealers' reactions
as, "Can't you automate this more? We want to access the information
immediately. It wasn't that they were against microfilm it was just that
they wanted even more automation."
In an effort related to the 1632 program, GM planned to create an auto-
mated parts catalog database. PSD was awarded a $1 million contract to
design the automated parts catalog data base for GM. The project was com-
pleted successfully and it generated a small profit.
A meeting was held in 1980 between GM and PSD in Flint, Michigan,
to conclude the negotiations for the 1632 rollfiche reader project. At that
meeting, GM's management told PSD management that the automotive in-
dustry had suffered a serious decline and that no more contracts could be
signed. Those close to the project noted that technical problems with the
1632 and the fact that it did not meet users' needs were as critical as the
recession to the program's cancellation. Write-offs on the program's can-
cellation totaled $3 million and total program losses exceeded $5 million.
In addition to this, during the early 1980s several events occurred that had
a serious impact on PSD's business.
Silver Crisis
The first shock to current business practices was the silver crisis. Silver
was the predominant film for parts catalog microfiche systems. Owing to its
higher quality, silver fiche could be left on a microfiche reader for extended
periods of time with no deterioration. When silver prices soared, PSD's silver
film business was dramatically affected. As the prices of silver kept rising,
so did silver film prices. The price of silver in August 1979 was $10 per
ounce..By December 1979 it had risen to $34 per ounce and reached an all-
time high of more than $50 per ounce by January 1980. John Marken recalls,
"the silver crisis drove us and our customers to diazo film."
Dave Gump, who was then general manager, remembers, "The profit
margin on silver was higher, and we had few competitors in silver microfiche.
The profit margins went down significantly with the switch to diazo film."
Don Prince, who is now vice president of operations, was production
manager at the time, and notes that with the switch from silver to diazo:
OCR for page 40
40
FREDERICK A. FELLOWES AND DONALD N. FREY
Many companies saw ours as an attractive business and for relatively small capital
investments were able to start small diazo microfilm businesses. That's not to say
that they could compete with us in terms of our copy preparation and film expertise,
but these smaller companies were capable of nibbling away at our business.
The result was that the microfiche service bureau business began to de-
teriorate into a commodity business.
Search for High Tech
Don Gardner, application development manager, recalls that during this
period, a second factor began affecting PSD's business:
Our customers were increasing the sophistication of their publishing. Industry was
waiting anxiously for the birth of office automation. The panacea of the future was
in a black box new tech was coming and our customers believed it was going to
make our microfilm outdated. This was making our business extremely difficult. We
tried to prepare to meet the technological opportunity.
As Dave Gump recalls, "the business was experiencing declining mar-
gins we were the only source of new business for our competitors. The
company had to find a way to change the product or expand the business."
In April 1981 PSD ventured into prepress services by acquiring DDSI,
which offered high-speed plotting of computer-generated illustrations, high-
speed typesetting, and data base publishing services. PSD's goal was to
vertically integrate its publishing services by moving upstream into processes
done before micropublishing. Dave Gump's comments on the DDSI acqui-
sition:
Automated technical publishing (ATP) was a very complex business. Our microfiche
customers wanted in-house capability, not a service bureau. We thought the De-
partment of Defense would be an excellent market for ATP. As we found out, there
was no incentive for DOD contractors to reduce their labor since they were cost-
plus. In September 1982 we closed the California office (losses totaled $800,000)
and brought ATP back to our Wooster, Ohio, facility. We are still selling ATP
services.
Included in the acquisition was a very high resolution photocomposition system
that output paper and microfilm. To utilize the equipment, we learned to create
programs to handle customers' photocomposition files (before this, with the exception
of GM's photocomp files for the 1632 program, PSD processed only standard data
processing files). We became much beKer at talking with publishing departments.
This equipment made a new venture with GM possible.
A FRESH START
Microfiche Catalog GM II
In February 1981, PSD started development of a standard 105-mm micro-
fiche parts-catalog system designed for small-to-medium-sized GM dealer
OCR for page 41
PICTURES AND PARTS
41
ships. In 1982, PSD secured a contractual arrangement with GM for the
microfiche parts catalog program, because of the savings to GM in switching
from paper to fiche-the traditional microfilm benefits. Don Gardner ex-
plains:
We were able to design some user-oriented indices to make it acceptable. Our system
design approach has made us unique in this marketplace. We didn't just sell micro-
fiche, we designed information retrieval systems. We even designed and redesigned
fiche readers to meet the applications.
The microfiche program was presented to select GM dealerships in Houston
in April 1982 (see Figure 1~. These dealers indicated that the microfiche
system was acceptable, but they strongly preferred an automated solution.
They asked for "push button" access to the GM factory parts data; 67 percent
said that they would wait for a more automated solution. But 25 percent
were willing to switch to fiche from paper-enough for the program to
proceed.
John Ramagli, then vice president of sales, recalls:
In April of 1983, the first GM parts catalog microfiche publication went out to the
field. We had reasonable support from the GM dealers. The GM Fiche Program is
ongoing today with 1,500 to 2,000 of GM's 10,000 dealers. At the same time,
however, GM's management was strongly urging us to do something else.
FIGURE 1 A 1982 presentation to GM dealerships yielded information about dealers'
preferences on automation of catalog access.
OCR for page 42
42
FREDERICK A. FELLOWES AND DONALD N. FREY
A Hybrid Solution Pursued GM ill
By this time, PSD had made several attempts to improve its business.
Acquisition of DDSI, increased marketing efforts directed at traditional busi-
ness, and internal productivity incentive programs were all undertaken.
PSD's limited market had matured. New products to speed catalog access
were seen as necessary to preserve the customer base and ensure growth.
PSD marketing continued to hear parts catalog users demand "automated
access." At the same time, PSD's customers were in various stages of
automating their parts catalog data bases.
PSD marketing proposed development of a system that would use micro-
film for graphics and a PC for text. A hybrid solution, using microfilm for
parts illustrations, was proposed rather than an all electronic approach that
the market seemed to be seeking because (1) no computer solution that had
adequate storage space for all the parts catalog illustrations was available;
and (2) the cost per image in microfilm was stunningly cheap compared to
the alternatives.
John Marken remembers:
The standard fiche program was very important in keeping the door open to future
more automated solutions for the parts catalog. We kept talking with GM, updating
them. Within PSD, John Ramagli developed an idea for a more automated system
for GM dealers. The concept was to employ cartridges containing high-reduction
images of parts illustrations printed on high-density 35-mm strips. The PC would
assist in selecting the correct cartridge, strip and frame as well as providing access
to text portions of the catalog.
The concept was presented in April 1983 at the annual Bell & Howell
Technology Conference, an interdivisional forum to promote discussion of
technology trends and to present new product concepts. John Marken re-
members Donald Frey, the chairman of Bell & Howell, suggesting that PSD
should pursue an all electronic solution because dealers would not accept a
mixed microform/PC system. Dave Gump recalls that it was also at this
conference that he first learned about low-cost compact disk read-only mem-
ory (CD-ROM) drives. At this point CD-ROM technology was evolving and
interest in it was growing.
In response to PSD Marketing's desire to offer a more automated parts
catalog system to GM dealers, Don Gardner became involved. He explained:
In the summer of 1983 Dave Gump called me to discuss Marketing's proposal for
a system to improve the productivity of our GM fiche users. In order to prove
productivity improvement, I needed to know the current productivity. In July of 1983
we proposed that we would have a survey done. I hired an industrial engineer and
designed the measurement methods and in September we proceeded with on-site
studies of dealers in the truck/transportation, agricultural, heavy equipment, and
automotive industries. The focus was on the parts department and particularly on
OCR for page 43
PICTURES AND PARTS
43
parts catalog lockups. We took into consideration the size of dealerships. We did
time studies and analysis of their operations (including profit and loss and manage-
ment records) to understand the parts business and its impact on the dealership (by
industry).
The Electronic Parts Catalog (EPC) Concept GM IV
The research verified a product need, but not PSD's product concept.
Dealers would not realize sufficient benefits to justify a partially automated
solution in which a cartridge microfilm reader driven by a PC was priced at
$5,000 per workstation. The concept of PC-aided-retrieval of microfilm for
parts catalogs would no longer be pursued dealers insisted on a completely
electronic system. The question became: could the likely price of an all
electronic parts catalog (EPC) be supported?
Results from this initial study led Don Gardner to recommend that the best
opportunity for developing an EPC was GM because (1) the automotive
industry represented the best single market segment; (2) within the automotive
industry, GM had the largest dealer network; (3) GM had an automated parts
database that PSD had designed and implemented; and, (4) PSD had also
implemented a catalog fiche system for GM (gaining experience in processing
GM's photocomp tapes) and was already familiar with their parts lookup
. .
application.
In November 1983 Don Gardner began to concentrate application research
on the GM parts catalogs at GM dealerships. Detailed data on their parts
lookup activities were recorded. The research verified the need to automate
the parts counterperson's catalog lockups.
However, from Don Gardner's research (completed in March 1984) it was
determined that a parts counterperson uses the catalog in just under half of
all transactions. Actual use of the catalog accounts for approximately 10
percent of a parts counterperson's day. Just over a third of parts catalog
lockups involve the illustrations.
Dave Gump notes:
Reducing the parts department's labor cost for catalog lockups was what we expected
would provide the cost justification for an electronic catalog. But when Don dis-
covered that only 10 percent of a parts counterman's time is spent using the catalog
we almost killed the project there could be no way to justify the product. But we
also learned that 16 percent of the counterman's time was consumed in transcribing
part numbers and entering them in the parts inventory system.
By interfacing an EPC with the parts inventory system nearly all the
transcribing and key entry could be eliminated-possibly enough productivity
improvement to justify purchasing the product. The design specification was
modified to include a hardware interface to dealer computer systems and
software to communicate with the parts inventory program.
OCR for page 44
44
FREDERICK A. FELLOWES AND DONALD A<. FREY
GM Dealer Business Environment
At a dealership the parts business consists of three components: retail,
wholesale, and service. The retail parts business is usually small. The whole-
sale business is the sale of parts to repair shops, garages, and other dealer-
ships. The service business is parts consumption by the dealer's own service
operation.
GM has 50 parts catalogs covering 1976-1987 model cars and light trucks.
Each catalog is two catalogs in one: the white pages contain part name, part
number, and usage data; the green pages have illustrations and references to
the white pages. It generally takes 18 months for a parts counterperson to
become proficient at using the parts catalogs. Experienced personnel fre-
quently are very well compensated, but despite that, parts counter personnel
turnover is high.
Parts lookup is complicated by the infrequency that individual catalogs
are updated and the flood of loose corrections, notes, and bulletins that are
received between new editions of the catalogs. The process of inserting the
bulletins and finding the notes associated with particular parts often is too
cumbersome to actually be done.
It is not unusual for a dealership to return to GM each year more than 5
percent of its ordered parts, usually owing to mistaken part numbers. GM
offers a program of credits to encourage dealers to reduce returns.
Nearly half of all GM dealers have computerized inventory control sys-
tems, and most of the larger dealers have computerized inventory control,
usually as part of a "dealer services" system sold by a data processing
equipment value-added reseller. A value-added reseller typically combines
industry-specific software it has developed with purchased hardware and sells
the package directly to that industry.
The Challenge
PSD was confronted by a major task: to find a technical solution to au-
tomating the parts counterpersons' access to the GM parts catalogs in a
manner producing sufficient economic payoff. From Don Gardner's research,
the dealers had identified what they wanted in an automated system:
1. To access information in 1 second.
2. Illustrations and photographs to be clear and easy to see.
3. To integrate with existing dealer systems and be able to communicate text data
to their currently installed inventory management systems.
PSD was determined to find a solution for GM. John Marken explains,
"General Motors was the biggest company in our biggest market and they
had the biggest data base. We could see the opportunity. We had to hold
the focus on that one product."
OCR for page 45
PICTURES AND PARTS
45
At this point, the "beta" project team consisted of only Dave Gump and
Don Gardner. Don Gardner recalls, "In 1984 we began a year of 'airplane
research.' " They traveled around the country learning of the alternatives
available in CRT technology, high-volume electronic storage, data base man-
agement software, and image scanning.
Dave Gump explains his strategy for acquiring technology:
We didn't want to replicate what was already there. We wanted to use as many off-
the-shelf components as were available. We were also interested in finding solutions
to avoid sole source situations.
There were three initial technical issues facing the PSD "beta" project
team: data volume, response time, and electronic image display quality. Each
of these issues had to be resolved to design an effective electronic lookup
system. GM parts catalogs were extremely large paper-based data bases.
Each GM car line had several thousand illustrations and a similar number
of charts. In addition, a large single car line requires approximately 100
megabytes of storage for text data.
Response time, the second technical issue, was affected by data volume
and image resolution. Catalog users insisted on 1-second access and display.
Figure 2 shows the image data transfer time and the relationship between
resolution in dots per inch (DPI) and the seconds to display. As shown, the
higher the resolution, the greater the time for image display. Response time
considerations excluded standard approaches to networking and data de
compresslon.
The third technical issue, image display quality, centered on the counter-
person's desire to see detailed illustration drawings and to easily identify the
FIGURE 2 Image data transfer time versus resolution (via RS-232 link).
OCR for page 46
46
FREDERICK A. FELLOWES AND DONALD 1~1. FREY
part call-out number without "pan and zoom." Many illustrations in GM
parts catalogs contain very fine, complex detail. Also, because of PSD's
expertise in designing microfiche retrieval systems for service applications,
the project team was aware that service and repair documentation contained
photographs with very small print describing repair procedures (see Figures
3 and 41.
Sources of Information
In search of a technical solution for the "beta" project, Dave Gump and
Don Gardner began to investigate and evaluate the options available. They
2'' ""'2~2 ' in? . .......
. '- ''"'''"~''.'.2.,'2,' ,''
' . .,.'':.'? ?,'? ?~'~.$,:~.,,,''.'.
· '-' ?-:':: :.~"22
.,.,.,., ,:?, ?:::: ? >,?'~" . ................
' ,:,:2 ':,2'':2: ', :2.,., ::,2 , ': '' ,2 .' . ! - it. - :
:~: ::::: :::,:::: - :.:.::: :.: :.:.:.: :",2,22:"":" """' a,::::,? '' ,.'
'..:,'~',:' .."':'.' '.":.~.':":.:':,.2..:':..':""::'':..'.2:,'.:'.:".".~.'.':,.': :"':':':',
i: ::::: ::. :: ::: ::::::::: :$: .: :: .: ::: .:::::::: . :::::::::: :::::::::::
~2'"'' '.~"""' .:.$'2..'.'.,,,',~', ''.,.",, ':'', ..,,.,,. :... it........
''':' .:..'...:.~-.?,..'',,:2',',.,.,.~''' :'.,,:,....,' ,',..:.',?:':::,:'
'I' 'I ~ I'm ~ I' ' 2 " I??? 2 ': - - - : ' ' 2 ""'''" I:?" ?' 2 " "' ':
'aim'" "":'2' ?.?, ,. ',.,',',',, . : I?', ? '
:, ,.~: :,':.:',:': :: .:..:.. , ::::: ,2'' :.::':, :': ::..:.: :2:,. .':2:.:,:.:.:'., :,: :.::,::,
..... .. ..... a .2 ...............
.. , , , , , ,. 2Y??2.?. ? 2."'."' "" ,~/i
... 2'' 2.,,,,, ,. , ?' ,2' ?
i: . ,'':~': ,':',, .,.,,.,,. ,,.,,,.,,,,, ,,,~............................................
' ......
, ,,' , : ~? _, ?, ?',.' ' ''y. . .............
......... , ,.,., ? -.,
..... 2.,'' ''',:', ''.' ,.
:: :::::::::::.':, 2,'.,.'. :':,
'.'' 2 .. ,
'.,,:.'', ',
.................... ...
:,'. '.2,"', 2:', .....
,."''.' ".'.
' '2'
..... ' 22".',2,
.'.'.".2.2."".'.2.' .,.
.,:,:,: ' ",2 ,": ', ,. .,.
,2 2 2.' " ' '. .,.',
.. '.' ' ' ' ' ',.,.
':,.2,.' '.' ' :.' .,.
.2 ".'.2.2 '.'.' '.'.' ....
,',',',:... ,.::...'. '....', .', ....
............ ,., .,., ,,.,.,.,.
.''..'.'..'..'...'..2.'.,'. ..
.' ' '."'."".' " ,
,,.,::.. "":':'':' .''".""" ..
:' ':':::''"'''' '' ....
"""":'""":"'""""'"' ....
................................ ""''''
.................................. ......
:.: ::. .: .. :::~::
~ If.
...~....~..~. ..
~ . i.
:. ~ ..-: .:. ~
::::::::::::::
: ~
-
-
-
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :::::: :::::::::::::::: ::::::::::
..
:: ~ - . - .~ ... .
-~
"2:2:::2:'?_ 'I . ? ?? ' ' ' '" "'''' ''
.?` ?:.''?: 'it? "it. :??" ?' -:?X ~ ~ is?? '~?''~Y. .'Y>~?.
. ~_, a', ..?,,.',......
·~??~?~? ?*, ? ...., .~... ? :,:,:',
I"? ~ ? ? ~?"2 ' '""''
.? ~ ~ ,~:..~$' ?~ '~i' ~ ,?,?g,?,??.?~,?, i?,
I' - ?'~..~?. ~,?..,~?....
=',.,:.:: · :.:.:.; ·:: i. .
FIGURE 3 Illustration from GM parts catalog.
,., ~. ~
.'.' ...... ~
?? ~ ~
_ ~, 1
: ~ ~
. 2 .2.2 '2.2'.'~.2 Y,76: ?, ,~" it ~ ~ ~ i" "'it
-'"..
.........................
,???'.2'?? .' .'.2.~:2' " :: ''
·' ''' '.' . .' :????'
:,:.:,.,?,?,~.'-:.:,
-
OCR for page 47
PICTURES AND PARTS
FIGURE 4 Photograph of service manual.
carried sample graphics from one electronics company to another to be scanned
and displayed. They experimented with a number of 200 dots-per-inch (bi-
tonal) monitors, but the display quality was not acceptable. In early 1984
they met with a consultant affiliated with Massachusetts Institute of Tech-
nology (MIT), who was familiar with medical imaging and scanning systems
and found a satisfactory gray-scale imaging and display system. Other tech-
nology decisions were made. The specific relational data base was selected.
Evaluations and decisions on optical disk drives and high-capacity Winchester
disk drives were made. Incorporating a touch screen was considered, and a
source was selected.
By the spring of 1984 the team had defined what to do. In April 1984 the
project team presented its marketing and technical research along with its
product concept at the annual Bell & Howell Technology Conference. It
proposed building a prototype system. Subsequent discussions with senior
management focused on optical disk as the medium of the future, and thus
the direction for "beta" was set.
Dave Gump explains his thoughts on the importance of the Bell & Howell
Technology Conferences:
Not only did these conferences foster the sharing of technical research among the
Bell & Howell divisions, they also provided annual reinforcement and corporate
support for technological change. The conferences afforded us the opportunity to
present new product concepts which could easily have been stifled.
Organization Structure
At the time, Dave Gump was vice president and general manager of PSD.
Recognizing that the "beta" development program was going to be as de
OCR for page 48
48
FREDERICK A. FELLOWES AND DONALD N. FREY
mending as starting a new business, requiring full-time supervision, Dave
recommended an organizational change that John Marken approved. In May
1984 Dave Gump became the vice president of development responsible for
the program's research and development on a full-time basis.
Gray Scale
One key technology issue was image display quality and the decision to
use gray scale. Dave Gump explains:
We saw a demonstration that visually confirmed that gray scale solved our image
quality problem at an affordable cost. Our contact at a Boston consulting firm
explained that there was scientific support for gray scale and referred us to research
being done at MIT. This reassured us we were following a reasonable path.
Gray scale versus bitonal was an intense issue among the engineers in different
divisions within Bell & Howell. To prove that gray scale was the correct approach
to PSD's parts and service applications, we hired an outside consultant, Battelle
Institute, to research this issue.
Battelle confirmed that gray scale was the correct approach for our application.
Selection of gray scale (rather than bitonal image data) had the added
advantage of requiring only half the storage because Image Data Base (IDB)
used fewer lines of resolution. Despite these benefits and user preference for
gray scale, opposition to gray scale continued from engineers in divisions
that had adopted the facsimile (bitonal) standard for office imaging products.
Prototype Development
Don Gardner recalls the activities in 1984:
Interspersed throughout 1984 were meetings with GM dealer managers. 1984 was
not just a year of identifying hardware, it was a time to solidify the perception of
the system . . . redefine how the product would function, try our ideas of what it
could do, as well as listen to users' ideas on what they wanted, and then determine
what hardware was needed.
The expected benefits and estimated value of the IDB were constantly
being refined by market research. In addition to labor savings at the parts
counter identified earlier, it became clear that another benefit expected from
an electronic parts catalog was reduced part number errors. Error reductions
were expected because transcription could be eliminated, information was
more easily accessible, and publishing could become more frequent. As
result of fewer part number errors, customer satisfaction would be improved
and restocking fees could be much lower. It was also believed that IDB's
ease of use would reduce the lost time and productivity of new employees
learning the GM parts catalog.
By July 1984 the PSD project team had developed a full proposal on the
OCR for page 49
PICTURES AND PARTS
49
costs to design and build a prototype system. In August 1984 this proposal
was presented to the chairman on his annual visit. The funding necessary
for prototype development was estimated to be $425,000. "How could deal-
ers justify purchasing a $50,000 electronic catalog system if a $5,000 1632
reader could not be justified?" This was the central question posed by the
chairman. An economic justification based on parts counter productivity
gains, reduction in part order error rates, and enhanced customer satisfaction
revealed that at $10,000 per workstation, the EPC was justifiable whereas
the 1632 was not. The chairman agreed in principle, and remembers that
throughout August, PSD management proceeded to push the proposal up
through channels for funding approval "with acceptance at best neutral from
outside the division."
On September 6, 1984, an outside consultant was contracted to develop
the prototype. During the same time, the project team increased to five people.
In response to the growing need for prototype hardware development to be
carefully supervised, Dave Fehr was hired as manager of Systems Integration
in October. Two application programmers were hired to begin the software
design.
Throughout the fall and winter of 1984, Dave Fehr supervised "beta"
prototype development, including design and coding of publishing software
and workstation "retrieval" software. The software engineering effort ben-
efited substantially from PSD's participation (1978-1979) in the development
of GM parts catalog data base. The two "beta" programmers were assisted
by the original data base designers. The "beta" publishing software was
able to incorporate files and code from the 105-mm fiche publishing software
for GM parts catalogs. Months of design, surprises, and redesign owing to
complexities in the data were avoided as a result of experience from the 1632
and 105-mm fiche programs. By December the prototype system, consisting
of a display workstation and scanning system, was up and running in the
1984 GM Cavalier parts catalog.
In January 1985 Don Gardner developed a slide presentation that explained
how the system and parts lookup application would work, how it would be
used, and what would be delivered. Don explains:
We had our first monitor, and the first images displayed on the monitor were so
good, I was really impressed. We had a photographer develop a series of slides of
the display. We took our slide presentation to the GM dealers in Houston the same
dealers who back in 1982 had said they would wait for a fully automated system.
We got absolute confirmation that functionally the system was going to do more for
them than they had asked.
John Ramagli recalls:
On February 28, 1985, the first people to see a demonstration of the parts catalog
lookup application on the "beta" system were the managers from General Motors.
We flew them from Detroit into Wooster to see the prototype.
OCR for page 50
50
FREDERICK A. FELLOWES AND DONALD N. FREY
At this point all we had on the system was the 1984 Cavalier parts catalog. We
showed them a demo of the system. They loved it, and urged that PSD take it to
pilot. (GM provided no funding nor even a commitment to allow PSD to publish
the catalog data electronically on a commercial basis.)
In February and March, the prototype system was demonstrated to Chrysler and
Ford. Chrysler showed a reasonable level of interest. Ford's interest was modest.
During June and July the prototype (with only the 1984 Cavalier) was
demonstrated to GM dealers throughout the country. The PSD project team
confirmed that they had a viable product and through research were able to
establish their goals- a $10,000 workstation. In August PSD received fund-
ing for full development of the project. During the remainder of 1985 the
production and software development team built the data base for three GM
car lines: Chevrolet, Oldsmobile, and Light Truck. Marketing identified the
activities that would be required to bring the product to market, and the
"beta" product was named the Image Data Base 2000 (IDB20001. In addition
to development activities on the GM parts catalog application, during 1985
research projects on GM dealers' service and repair procedures and Chrysler
dealers' parts operations were started.
Field Tests
In January 1986 the IDB2000 prototype system was demonstrated at the
National Automotive Dealers Association (NADA) trade show in the booth
of a parts inventory system vendor specializing in very large "mega" dealers.
PSD's Marketing team successfully solicited pilot dealerships and support
from GM management. Several small companies demonstrated videodisk-
based systems with sample pages from parts catalogs. The image quality
was poor; vital reference numbers ("call outs") were illegible. Computerized
microfilm systems for parts catalogs were shown by a major vendor of data
processing systems to dealers and by others. The strongly positive reaction
to IDB convinced PSD to further accelerate the IDB program.
In March PSD and GM management held a review meeting to discuss the
criteria for a successful field test and the strategic implications of a good
pilot program. Lustine, the largest GM parts dealership, and Hutton Chev-
rolet, the second largest, were selected to be part of the pilot program. GM
requested that the pilot be expanded to include a medium-sized dealership,
Bell Dean, and that a system also be installed at the GM Technical Center.
In April field tests of the IDB2000 parts catalog application began. The
first field test location had only three workstations, and yet the users com-
plained about the system's response time. Internal testing with many more
workstations (large customers would need 15 to 25) revealed that commu-
nications completely bogged down in the CPU, largely because of the com-
bined overhead and interaction of the network software, the relational data
OCR for page 51
PICTURES AND PARTS
51
base, and the operating system. It was a major setback. In June after much
consideration, the PSD development team determined that IDB2000 design
changes would be necessary. Dave Gump explains:
The number of users that we had to support and the performance requirements
indicated that it would require a very large computer. Our target price couldn't
support that size computer. We chose to put the application into the user-level
computer, eliminate the CPU, and switch to a network server for data base storage.
John Ramagli remembers his discussions with GM management about the
design revision, which had been named the "Jade" project:
It was a crisis. At this point we had growing concerns about the speed implications
of using the MicroVAX II computer to run the application on a multi-user workstation
system. I met with the director at General Motors to explain to him that our decision
to design and build our own hardware would significantly reduce the cost and increase
the performance.
GM agreed to have PSD continue with the pilot tests, and, as part of the
program, PSD agreed to replace each of the pilot sites with Jade systems.
In July 1986 the design revision was funded and the Jade project began. The
first Jade system was available in October and the system was installed at
the Lustine dealership. Lustine's management was so pleased with its im-
proved performance that it offered to buy the system. On November 19,
1986, after completing his pilot test research, Don Gardner returned from
Lustine with the first purchase order.
As Jade systems were installed at the remaining GM pilot sites, the mar-
keting group was developing the product introduction plans for the IDB2000
EPC and working with GM to complete the negotiations for rights to the
parts catalog data base. On December 18 the contract providing Bell & Howell
with rights to the data base was approved by GM. PSD worked with several
units of GM including, for data processing, Electronic Data Systems, a billion
dollar provider of data processing services and automation acquired by GM
in October 1984. The diverse interests of each unit made securing a com-
mitment from GM an exacting process.
During the field test many recommendations were made by the users for
improvements in the application software, but of even more significance was
how the system was used. One dealer placed a workstation in the service
area (rather than placing all at the parts counter), and discovered that me-
chanics could find their own parts and have the list of requested parts printed
out in the stock room. Each mechanic was thus able to increase billable time
an average of 5 hours per week. The additional service work provided more
financial justification than parts counter labor savings, reduced part number
errors, or any other quantified factor. Table 1 summarizes the measured
productivity improvements owing to IDB installation.
Although none of the four dealers that field tested the Jade system was
OCR for page 52
52
FREDERICK A. FELLOWES AND DONALD N. FREY
TABLE 1 Benefits of Image Data Base for Dealers
MEASURED PRODUCTIVITY IMPROVEMENTS
Wholesale parts orders/day: Up 20% after image data base installed.
Service revenues: 10% increase if mechanics order parts; 2% increase otherwise.
Parts returns: Reduced from 7 to 6% without emulation. Reduced from 7 to 4% with
emulation (dollar-for-dollar credit for reductions paid by GM to dealer).
DEALER CHARACTERISTICSa
Annual wholesale parts: $1,940,000 (gross margin 16%)
Annual parts sales to service: $669,327 (gross margin 22%)
Annual service labor revenues: $2,430,000 (gross margin 54%)
aProspect for 10-workstation ($100K plus 20% per year operating costs) system. A dealer of
half the size would need a 5-workstation system. This "ideal" dealer is evenly split between
wholesale parts and service; most divisions are skewed in favor of one or the other.
interfaced to inventory control systems, all four purchased systems. It ap-
peared that interfacing to the inventory control system was desirable but not
required by the market. Lack of cooperation from the two major inventory
control system vendors was likely to turn into opposition and competition at
product launch.
Product Launch
Early in 1987, PSD's Marketing and Sales Management was completing
the plans for the IDB2000 product introduction and adding the marketing
and sales staff required to successfully launch and actively sell the system
to GM dealers. A training department was established and customer support
specialists were hired to be prepared to train IDB2000 customers. Sales
managers and a national sales.force were hired and trained. The Bell &
Howell national service force was trained on installation and repair procedures
for the IDB2000.
In February 1987 the IDB2000 EPC was introduced at the NADA trade
show. Figure 5 shows the terminal and the display quality in the commercially
available EPC. The IDB was in five booths: GM, Chrysler, and three vendors
of dealer data processing equipment. The IDB and competitive EPCs drew
large crowds. The two largest data processing equipment vendors and IBM
demonstrated PC-based EPCs that did not use gray scale. Several publishers
that were developing CD-ROM products for libraries presented parts catalog
concepts at other shows the next month. Computerized film retrieval and
videodisk parts catalogs were absent.
In March sales of the IDB2000 began, and during this month the first
"nonpilot" system was sold. Although no competitive electronic GM cat-
alogs were on the market, a vendor of data processing equipment to dealers
OCR for page 53
PICTURES AND PARTS
53
_-1 ~
.... ~
~ ~ ~ ~ ~ ~ 1 ~ ~ ~
~ , _ - ~ ~ l e. _
FIGURE 5 Terminal for IDB2000 electronic parts catalog.
was field testing an Acura EPC with Honda dealers. The two largest data
processing equipment vendors were advising dealers not to buy the IDB, but
sales continued to be strong-surpassing one large system sold per day within
3 months of launch. During May the Chrysler Parts and Service Technical
Information System pilot program began that included pricing, supersedence
and service bulletin data bases not included in the GM application.
By the March launch PSD had spent $11 million on development (both
engineering and marketing) of the IDB System. When added to the 1632
project, the loss on DDSI, and the 35-mm cartridge film/PC study, total
investment approached $18 million. This was also the amount budgeted for
IDB Sales in its first year and nearly the annual sales for the entire division
in past years for micropublishing.
From the point of launch, major cost reduction redesigns were under way
to expand the target market to include medium- and smaller-sized dealerships.
What brought the IDB2000 this far? Dave Gump comments:
First, the IDB2000 is really an extension of current business. We built on what we
OCR for page 54
54
FREDERICK A. FELLOWES AND DONALD N. FREY
know. Our forte is application expertise and the design of information retrieval
systems for technical documentation users. We have combined technology in an
innovative way to meet our customers' needs. Bell & Howell Company has always
had a desire and an attitude to move the company technologically forward.
John Marken attributes IDB2000's success to perseverance. He recalls:
We could have easily walked away from pursuing "beta" but we were determined,
and the risk was worth the effort. Back as far as the late 1970s, the dealers knew
what they wanted. The technology just didn't exist to provide them with fast access
to pictorial part information at the right price.
We have been successful because we have been able to combine talent and knowl-
edge and have always been focused on what our customers wanted. It took guts and
sponsorship. Everybody we reported to wasn't necessarily in favor of it. But, we
had a sense of customer need and corporate support.
By dwelling on the needs of our customers and being persistent in our search for
a solution we have met with success in the IDB2000. At all times our strength was
our focus on the application, the dealers' needs, and the ability to include the right
technology to meet their needs. It is no question that we were living on the forward
edge of technology and continue to do so.
April to October 1987
During the spring and summer of 1987 a steady stream of software changes
that improved access, saved storage space, or made system usage easier
(particularly updates) were successfully released to users. The fine-tuned
application software incorporated an understanding of GM's data and its use
that PSD had gained over the past decade. Also, a substantially lower cost
"pedestal" was developed to replace the data storage "rack." The new
device allowed affordable pricing of smaller systems (two to four work-
stations).
The major dealer computer systems (DCS) vendors' efforts to discourage
IDB sales became more effective over time. By midsummer IDB sales pla-
teaued as the DCS vendors promised full integration of their future EPCs to
existing packages (such as inventory control packages), and demand for
interfacing the IDB grew. In August a nonexclusive, sole distributorship
agreement was reached between PSD and the largest DCS vendor, Reynolds
& Reynolds. The second largest DCS vendor announced a competitive prod-
uct and acquired access to the GM parts catalog data.
PSD developed simple interfaces to the more popular DCS packages. But
without support from the DCS vendor the simple interfaces were awkward
at best and often unworkable. In response PSD developed "emulation"
software. This software allowed the IDB2000 workstations, when intercon-
nected to the DCS system, to behave exactly as the DCS vendor's terminals
did (i.e., emulation). By October the emulation package had become very
OCR for page 55
PICTURES AND PARTS
55
popular and was displacing DCS terminals in dealership parts departments.
Emulation was expected to create more value to the dealers than simple
interfacing. In addition to increased speed and accuracy no keypunch or
transcription errors provided by simple interfacing, emulation was expected
to be more convenient and save counter space, equipment cost, and time
spent moving from IDB workstation to DCS terminal. By November nearly
all IDB System sales included purchase of emulation software.
At the end of October 1987, PSD's fiscal years revenues were forecast to
achieve 100 percent of budget and 1988 IDB revenues were budgeted at
approximately twice those of 1987.
CONCLUSIONS
This case chronicles the efforts of Bell & Howell PSD to meet its cus-
tomers' desire for an "automated parts catalog lookup system." It describes
the transformation of a micropublishing service bureau through the search
for a technical solution to meet its customers' unique information retrieval
requirements.
The EPC system developed is a back-room system for GM auto dealers.
It is one part of the dealer's strategic information system. Its value is largely
owing to its effectiveness rather than its efficiency. Originally envisioned to
increase efficiency by reducing parts catalog lookup time, and then to reduce
part order errors, the EPC has been found by dealers to increase service
billings (labor and parts) up to 15 percent providing for more profit op-
portunity than possible from cost savings.
This case presents the trends and issues affecting the micropublishing
industry as far back as the 1970s, and identifies specific events that led
management to pursue the development of the IDB2000. Five management
and policy issues were critical to the success of the IDB2000 and responsible
for PSD's ability to achieve technological change.
The first and most important issue is the division's constant focus on its
customers' needs. Study of users continued through the field test and into
product launch, revealing new insights well after the formal studies ended.
Efforts by PSD to understand the parts business and its impact on dealerships
provided confirmation of the product need and a clear understanding of the
user benefits to be achieved through "automation." By performing detailed
time studies, PSD gained a complete understanding of the parts counterper-
son's daily use of the paper parts catalog. This provided the foundation for
PSD to develop the parts lookup application based on how the catalog user
wanted to access part description, part number, and part illustration infor-
mation. Ongoing customer surveys, focus groups, and one-on-one interviews
were responsible for the achievement of a system that has received excep-
tionally positive dealer responses.
OCR for page 56
56
FREDERICK A. FELLOWES AND DONALD N. FREY
The second central issue is PSD's willingness to apply new technologies
in developing a solution to market needs. The task of evolving from microfilm
to electronics technologies was recognized as difficult and risky, but the
alternative was a shrinking business. With a knowledge of the user's re-
quirements for information retrieval, the PSD project team was able to iden-
tify the technical issues essential to providing dealers with a fully electronic
solution. The team's dedication to creating a solution the market desired,
including 1-second image access and display, as well as high-quality display
of parts illustrations, resulted in the pursuit of system components (including
infrared touch screens, gray-scale monitors, optical-disk storage devices, and
networking) that were on the leading edge of technology.
The third factor was management's persistence for 10 years from the ill-
fated 1632 project to the successful launch of the IDB2000. Despite several
failures and mounting costs, the convictions of a few people carried the
program forward. Dedication to the final goal was coupled with an open
mind toward implementation. Engineering plans and the perceived value to
the user both changed substantially more than once. Management was ded-
icated to success but flexible enough to change quickly in response to constant
market feedback and engineering experience.
The fourth factor in the success of the IDB2000 project was the corporate
desire to move the company forward. A cultural direction favoring change
was emerging. Sponsorship of annual interdivisional Technology Confer-
ences provided a forum to promote the sharing of technology trends and
presentation of new product concepts. Through attendance at these confer-
ences, PSD's management gained valuable information about new technol-
ogies, was also able to present its product ideas, and gained ongoing corporate
support and funding for the IDB2000 project.
Last, strong project and program management was critical to the successful
development of an EPC. PSD had little experience with developing electronic
equipment, and the technical resource requirements were many times larger
than previous division staffing. PSD reorganized for a program of these
proportions, locating the entire team in nearby but separate facilities. Classic
project management processes were implemented and refined to specific
needs. As a result, the program manager could more effectively manage and
control PSD's efforts to develop an EPC while effectively and credibly
reporting progress and problems to management outside PSD.
Representative terms from entire chapter:
gray scale
