You are on page 1of 45

AL5O IN5I0: AL5O IN5I0:

NAkCR/ APkl | 2010 DklVlNG BETTEk BUSlNESS DEClSlONS


- FINANCIAL FIA5CO: |olly ol fnonciol modclcrs
- 5OLUIION 5IAkIUP5: Scvcn principlcs ol succcss
- VHICL kOUIING: Survcy ol culling-cdgc sollworc
R T T P : / / W W W. A N A | Y T l C S N AGA Z l N E . C O N
kVNU MANAGMNI

V5.
5OCIAL NIWOkKING
Counlcr-mcosurcs
oimcd ol lnlcrncl-
conncclcd cuslomcrs
Excculivc Edgc:
Cisco V.P.
Kcvin Rorringlon
on voluc cboin
monogcmcnl
* u : G: ER M B <L F:@: S B G> ' <HF
NOTREkS O| lNVENTlON
l NS l DE S T OkY
If necessity is the mother of inven-
tion, then deregulation is the father,
and revenue management is the
couples golden child, at least as far
as the airline industry is concerned.
The Airline Deregulation Act of
1978 spawned a host of low-fare car-
riers that could make money while
selling seats at a fraction of the price
charged by established airlines. That
gave the upstarts a huge advantage.
The big carriers, locked into hefty
labor and legacy costs, had no clear
way of leveling the playing eld. Their
sheer size became a liability. All ma-
jor airlines faced a troubling question:
How can we compete? If we lower
fares across the board, it doesnt mat-
ter if we have a 100-percent load fac-
tor well still lose money. And if we
dont match the upstart airlines fares,
well lose market share, and eventu-
ally well be out of business.
American Airlines solved the dilem-
ma by offering different fare classes in
each market. American matched the
low-fare airlines prices with a portion
of its seat inventory on certain routes
and saved another portion for late-ar-
riving, high-yielding demand (typically
business travelers).
Thus was born the concept of rev-
enue management, but several ques-
tions remained: How many seats do
you sell at a discount and how many
do you hold in reserve? When and by
how much do you change the price
of each bucket of inventory? Again,
analysts and their tools of the trade
forecasting, optimization, etc. came
up with the answers.
Virtually every airline in the world
adopted some sort of RM system,
and the concept inevitably spread to
many other industries (hotel, cruise,
etc.) where you nd perishable,
limited capacity. Alas, RM couldnt
solve every corporations problems.
Many airlines still went out of busi-
ness, crushed by some combina-
tion of high legacy costs, soaring
fuel prices, post-9/11 paranoia, the
recession and good, old-fashioned
corporate mismanagement.
Today, the purveyors and man-
agers of RM systems face a whole
new threat: the Internet and the
tech-savvy customers it spawns as
outlined in our cover story.
- PETEk ROkNEk, EDlTOk
borncrlionbrlpub.com
PAkAGON
|U|| PAGE AkTWOkK
G>P:KMPHKD
PAGE 1
Experience how AIMMS
enables you to apply
OR successfully
AIMMS is a registered trademark of Paragon Decision Technology B.V.
Other brands and their products are trademarks of their respective holders.
Europe, Middle East & Africa = Haarlem, The Netherlands = Phone: +31 (23) 5511 512
Americas = Kirkland, WA, USA = Phone: +1 (425) 576 4060
Asia Pacific = Singapore = Phone: +65 9640 4182
www.aimms.com = info@aimms.com
Transparent and maintainable
model formulation
Point-and-click integration
with data sources
Seamless links to
powerful solvers
Fast GUI builder to visualize
results and to interact with
your model
Hassle-free deployment
alternatives to end-users
lree AlMMS Viewer
We/ Services
lxcel aJJ-in
COM o/ject
All
etc.
and many
more...
ILOG CPLEX
CONOPT
OLE DB
OBDC
...
Please visit our website for application
examples, industry case studies and a
free full trial license.
www.aimms.com
DRIVING BETTER BUSINESS DECISIONS
C C | | | S
FAIUk5
kEVENUE NANAGENENT VS. SOClA| NETWOkKlNG
;rRnkbE^obgZg]C^__F\@bee
/| c |.c w|c |cc|scvvy \cccccc|cc ccscs cc
cc|c || c|s |cw s|c|c cccc.cs csccc
SUCCESS|U| SO|UTlON STAkTUPS
;rC^kkr;ZgdlZg]D^gFnll^efZg
Scvc scs|.cs cvcy sys|cs cc|ys| cccs |c 'cw .
ccc |c |cc| c cc,cc| 'c c cccc' c ||c .|| cc|
ON TRE kOAD TO CONNECTlVlTY
;rCZgb\^IZkmrdZZg]KZg]heiaAZee
\c|.c|c c|. sc|wcc svcy ccc|.vc .|cc|.c c
ccc|c cc.cc|.c cc |ccc|.c |cc|c|cy
ANA|YTlCS AT UPS
;rC^__Pbgm^klZg]=hn`Fhak
Ccccc|c cc||c ccv.ccs c .s.cc |cc' c| ||c |cc| |cc|s
cccwc. $' c.||.c cc|.vcy cc |c.s|.cs ccccy
WOk|D-C|ASS CON|EkENCE
;rMa^k^lZ<krZg
/cc|y. Sc.ccc |c ||c /| c |s.css |cv.cw c
|||C||S s|c'c cc|y|.cs cccccc . C|ccc
12
17
22
29
32
+ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u PB GM > K +) )1
Easier, More Powerful and Far More Visual.
From its classic Solver Parameters dialog to its modern
Ribbon, modeless Task Pane, and charts to visualize the
and produce instant reports, charts and graphs, without
programming. And you can do even more in Excel VBA.
and a wide array of charts and graphs. One simple,
consistent user interface makes learning and teaching
Free Trials and Support for Your Eorts at Solver.com.
about the many other ways we can support your
teaching and learning eorts all at www.solver.com.
THE BEST WAY TO TEACH AND LEARN OPTlMlZATlON
RlSK SOLVER PLATFORM
Tel 775 831 0300Fax 775 831 0314nfo@solver.com
32
22
12
|/|C|/ /||| | 2!
DRIVING BETTER BUSINESS DECISIONS
C C | | | S
*)%896)7
K>O>GN>F:G:@>F>GMOL'LH<B:EG>MPHKDBG@
;rRnkbE^obgZg]C^__F\@bee
/| c |.c w|c |cc|scvvy \cccccc|cc ccscs cc
cc|c || c|s |cw s|c|c cccc.cs csccc
LN<<>LL?NELHENMBHGLM:KMNIL
;rC^kkr;ZgdlZg]D^gFnll^efZg
Scvc scs|.cs cvcy sys|cs cc|ys| cccs |c 'cw .
ccc |c |cc| c cc,cc| 'c c cccc' c ||c .|| cc|
HGMA>KH:=MH<HGG><MBOBMR
;rCZgb\^IZkmrdZZg]KZg]heiaAZee
\c|.c|c c|. sc|wcc svcy ccc|.vc .|cc|.c c
ccc|c cc.cc|.c cc |ccc|.c |cc|c|cy
:G:ERMB<L:MNIL
;rC^__Pbgm^klZg]=hn`Fhak
Ccccc|c cc||c ccv.ccs c .s.cc |cc' c| ||c |cc| |cc|s
cccwc. $' c.||.c cc|.vcy cc |c.s|.cs ccccy
PHKE=&<E:LL<HG?>K>G<>
;rM^k^lZ<krZg
/cc|y. Sc.ccc |c ||c /| c |s.css |cv.cw c
|||C||S s|c'c cc|y|.cs cccccc . C|ccc
12
17
22
29
32
+ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u PB GM > K +) )1
Easier, More Powerful and Far More Visual.
From its classic Solver Parameters dialog to its modern
Ribbon, modeless Task Pane, and charts to visualize the
and produce instant reports, charts and graphs, without
programming. And you can do even more in Excel VBA.
and a wide array of charts and graphs. One simple,
consistent user interface makes learning and teaching
Free Trials and Support for Your Eorts at Solver.com.
about the many other ways we can support your
teaching and learning eorts all at www.solver.com.
THE BEST WAY TO TEACH AND LEARN OPTlMlZATlON
RlSK SOLVER PLATFORM
Tel 775 831 0300Fax 775 831 0314nfo@solver.com
32
22
12
|/|C|/ /||| | 2!
Analytics (ISSN 1938-1697) is published six times a year by
the Institute for Operations Research and the Management
Sciences (INFORMS). For a free subscription, register at http://
analytics.informs.org. Address other correspondence to the
editor, Peter Horner, horner@lionhrtpub.com. The opinions
expressed in Analytics are those of the authors, and do not
necessarily reflect the opinions of INFORMS, its officers,
Lionheart Publishing Inc. or the editorial staff of Analytics.
Analytics copyright 2010 by the Institute for Operations
Research and the Management Sciences. All rights reserved.
|kONT|lNE
|U|| PAGE AkTWOkK
G>P:KMPHKD
PAGE 3
DRIVING BETTER BUSINESS DECISIONS
0PAkIMNI5
lNSlDE STOkY
EXECUTlVE EDGE
PkO|lT CENTEk
VlEWPOlNT
NEWSNAKEkS
ANA|YTlCS lN ACTlON
|AST WOkD
1
4
6
8
10
35
37
, u : G: ER M B <L F:@: S B G> ' <HF
REGISTER FOR A FREE SUBSCRIPTION:
http://analytics.informs.org
INFORMS BOARD OF DIRECTORS
President Susan L. Albin, Rutgers University
President-Elect Rina Schneur, Verizon Network & Tech.
Past President Don M. Kleinmuntz, Strata Dec.Tech.
Secretary Anton J. Kleywegt, Georgia Tech
Treasurer Stephen M. Robinson, Univ. of Wisconsin
Vice President-Meetings Robin Lougee-Heimer, IBM Research
Vice President-Publications Terry P. Harrison, Penn State University
Vice President-
Sections and Societies Ariela Sofer, George Mason University
Vice President-
Information Technology Warren Lieberman, Veritech Solutions
Vice President-Practice Activities R. John Milne, IBM
Vice President-International Activities Jionghua Judy Jin, Univ. of Michigan
Vice President-Membership
and Professional Recognition Pinar Keskinocak, Georgia Tech
Vice President-Education Jayashankar M. Swaminathan, UNC
Vice President-
Marketing and Outreach Anne G. Robinson, Cisco Systems Inc.
Vice President-Chapters/Fora Stefan Karisch, Jeppesen
INFORMS OFFICES
www.informs.org Tel: 1-800-4INFORMS

Executive Director Mark G. Doherty
Marketing Director Gary Bennett
Communications Director Barry List

Corporate, Member, INFORMS (Maryland)
Publications and 7240 Parkway Drive, Suite 300
Subdivision Services Hanover, MD 21076 USA
Tel.: 443.757.3500
E-mail: informs@informs.org
Meetings Services INFORMS (Rhode Island)
12 Breakneck Hill Road, Suite 102
Lincoln, RI 02865 USA
Tel.: 401.722.2595
E-mail: meetings@informs.org
ANALYTICS EDITORIAL AND ADVERTISING
Lionheart Publishing Inc., 506 Roswell Street, Suite 220, Marietta, GA 30060 USA
Tel.: 770.431.0867 Fax: 770.432.6969
President & Advertising Sales John Llewellyn
Llewellyn@lionhrtpub.com
Tel.: 770.431.0867, ext.209
Editor Peter R. Horner
horner@lionhrtpub.com
Tel.: 770.587.3172
Senior Art Director Alan Brubaker
albrubaker@lionhrtpub.com
Tel.: 770.431.0867, ext.218
Junior Art Director Kat Wong
katwong@lionhrtpub.com
Tel.: 770.431.0867, ext.223
Advertising Sales Sharon Baker
sharonb@lionhrtpub.com
Tel.: 813.852.9942
4
35
Risk Analysis is Easy as What-If with our Ultra-Fast,
In true Excel 2007 style, select from galleries, drag and
watch charts update instantly with new results.
produce instant reports plus standard, Overlay, Trend
and Box-Whisker charts, without programming.
the
consistent user interface makes learning and teaching
Free Trials and Support for Your Eorts at Solver.com.
about the many other ways we can support your
teaching and learning eorts all at www.solver.com.
THE BEST WAY TO TEACH AND LEARN SlMULATlON
RlSK SOLVER PLATFORM
Tel 775 831 0300 Fax 775 831 0314 info@solver.com
Managing a global value chain for a $40 bil-
lion company is a complex challenge. Thank-
fully we have 9,000 colleagues to help us in
our Customer Value Chain Management orga-
nization at Cisco. My role is to lead a glob-
al business operations team responsible for,
among other things, strategy and planning,
supply chain capability creation, ERP evolu-
tion and business intelligence across an end-
to-end spectrum of disciplines that we call the
customer value chain. The value chain, as we
dene it, goes beyond traditional supply chain
management and covers everything from new
product introduction and demand manage-
ment through order management, sourcing,
manufacturing, quality management, logistics,
recycling and re-use.
We developed this holistic approach as
part of a business transformation program
launched at Cisco approximately ve years
ago. Our strategy was to bring together for-
merly silo-ed teams around a shared nucleus
of governance, metrics and culture. The accel-
erated collaboration produced changes that
have worked well for Cisco and our customers,
helping us expand into new markets around the
world while improving the customer experience
and key metrics such as quality yields, on-time
shipment and inventory turns. Today, when my
boss asks for ideas on how to take our trans-
formation to the next level of productivity and
customer service, I begin my answer with two
words: operations research (O.R.).
O.R., I believe, presents us with key op-
portunities to up-skill our organization, conrm
our strategic directions and make better op-
erational decisions every minute of every day.
This has become clear, as our aspirations as
a company have broadened beyond our core
strength in networking equipment to encom-
pass as many as 30 new market adjacencies.
Cisco is not only expanding into new data cen-
ter, virtualization and video markets; we are
also reaching whole new customer segments
such as home consumers making their own
videos and downloading entertainment.
These new markets produce added com-
plexity. Working with outsourced contract
manufacturers and thousands of suppliers,
we now design and build more than 300 ma-
jor product families. Most of these products
are custom-congured and built-to-order, but
many now are built-to-stock. Some products
need carrier-grade reliability and high-touch
customer support. Others need to be localized
or manufactured and shipped within legal, reg-
ulatory and tax frameworks specic to different
countries, including fast-growing but uniquely
demanding markets in places such as China,
India, Mexico and Brazil.
In this multi-faceted environment, we no
longer manage the supply chain. Instead, we
are tasked to design and manage multiple
high-performance global supply chains with
different characteristics for multiple business
models. As the mission grows more complex,
and as the interdependencies among different
areas of our operations continue to increase,
it is becoming obvious that out-of-the-box soft-
ware and conventional program management
are not going to cut it. Increasingly, we are go-
ing to need hardcore statistical analyses and
other advanced analytics techniques to make
smart decisions on forecasting, risk manage-
ment, capacity planning and many other areas
across our global operations.
This optimism about O.R. is more than
just academic speculation. O.R. has already
made a significant impact on our value chain
in areas such as demand management.
Forecasting customer demand is, of course,
a central part of supply chain management
and a critical enabler of lean manufacturing.
This discipline becomes ever more challeng-
ing in times like our own characterized by
rapid changes in the macro-economy and
volatile swings in supply and demand. In fact,
Ciscos need to write off some unused inven-
tory after the dotcom bust in 2001 provided
some of the impetus for the larger transfor-
mation of our value chain.
Ciscos determination not to repeat that
experience led to an initiative to establish a
world-class demand management and plan-
ning capability. Constant changes in demand
and our complex product mix made it clear that
we couldnt count on the forecasting models
- u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
SEElNG TRE |UTUkE
lN VA|UE CRAlN
NANAGENENT
BY KEVlN RAkklNGTON
ClSCO`S STkATEGY:
BklNG TOGETREk
|OkNEk|Y Sl|O-ED
TEANS AkOUND A
SRAkED NUC|EUS
O| GOVEkNANCE,
NETklCS AND CU|TUkE.
E XE C U T l VE E D G E
in conventional demand-planning applica-
tions. In response, we began to recruit O.R.
experts and point them at the problem. They
responded by developing a forecast genera-
tion process that combines advanced ana-
lytics, which are applied to enterprise data
on bookings and historical demand, with a
consensus process incorporating current in-
sights from sales, marketing and finance. To
generate the best possible forecast for each
product line, the analytic process compares
the performance of a range of methodolo-
gies over the available data. Because de-
mand varies widely across different product
lines, this ability to tailor both the data set
and the analytic methodology has enabled
Cisco to achieve significant improvements
over a one size fits all approach.
The results include better forecast ac-
curacy, increased inventory turns and an
overall improvement in supply-demand bal-
ancing that has paid off for both Cisco and
our customers in the form of reduced excess
inventory and faster, more reliable service.
During the worst of the recent economic
downturn, Cisco was able to reduce inven-
tory in the supply chain without write-offs
or a fall-off in customer service. Today, our
statistical forecasting experts are working to
further refine the entire process and manage
the increased demand caused by the global
economic recovery.
Managing risks associated with the econ-
omy, political events, natural disasters and
other potential disruptions is another area
where O.R. expertise is helping us trans-
form our value chain. Analytics allow our risk
management team to quantify the impact of
a supply chain node being disrupted. By cor-
relating business continuity planning data,
probability data and ERP data, we can fo-
cus on those nodes that represent the high-
est risks to a supply chain. This helps guide
our capacity planning as well as contingency
plans, and we gain the ability to keep prod-
uct flowing to our customers independent of
world events. For example, when Bangkoks
airport was shut down by protestors in 2008,
Cisco had truck convoys ready to move
product from our partners nearby factory to
an airport in Malaysia, sparing our custom-
ers any disruption.
Looking ahead, many additional opportuni-
ties for O.R. exist within our value chain op-
erations. For example, we have made good
progress in implementing Six Sigma methods
across our contract manufacturing and supply
network, but we now have an opportunity to
analyze the results of those changes to rene
our future best practices and calibrate our in-
vestments. In addition, we are looking at ways
to use analytics to integrate resiliency into our
value chain design process. This would help
us balance risks with competing objectives
such as cost and exibility.
Overall, O.R. has the potential to enable
a new level of productivity in Ciscos global
value chain organization, along with an unri-
valed customer experience. Analytics enable
us to understand and automate an expand-
ing list of data-intensive tasks and quickly get
good answers to questions that are becoming
more complex every day. How, for example,
can we optimize innovation, operational ex-
cellence and customer service, driving down
costs even as we make new strategic invest-
ments? With the added insights generated
by O.R., we can secure competitive advan-
tage while accelerating performance. That is
a formula that pays off for our customers and
our business.
Kevin Harrington is vice president of Global Business
Operations in Ciscos Customer Value Chain Management
organization.
. u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
DUklNG TRE WOkST
O| TRE kECENT
ECONONlC DOWNTUkN,
ClSCO WAS AB|E TO
kEDUCE lNVENTOkY
lN TRE SUPP|Y CRAlN
WlTROUT WklTE-O||S
Ok A |A||-O|| lN
CUSTONEk SEkVlCE.
E XE C U T l VE E D G E
Ciscos global value chain includes fast-growing but uniquely demanding markets
such as China, India, Mexico and Brazil.
A common question when starting down the
path of analytics is whether to start big, small or
somewhere in between. Like so many things in
life, theres no one simple answer. There are,
however, some useful guidelines.
5IAkIING 5MALL
Common wisdom is to start small: find
a good, specific project, make it a success
and then build on that success. Management
at a waste management firm might ask if it
was possible to reduce the number of trucks
and drivers it uses through better dispatch-
ing. The problem can be examined by an
individual with a combination of operations
research and computer skills and a project
sponsor with the clout to get access to the
necessary data and people.
Starting small can be an effective approach,
but in practice it often indicates a lack of cor-
porate interest. If the project sponsor is the
CEO, the projects on solid ground. However,
if the sponsor is a mid-level manager with a
new idea, the project has a good chance of
going nowhere.
When starting small, its necessary to have
a clear plan for expanding. Project participants
need to spark excitement in the people they
interact with, both before and during the proj-
ect. Its also necessary to get management
commitment up front for moving forward if the
project is successful. If the commitment is,
OK, give it a try, and well look at the results
and think about what to do next, the projects
almost certainly not worth the time to begin
with. If the commitment is, OK, give it a try,
and if its successful, Ill help you get support
to roll out the results in the Eastern region,
the project has a ghting chance. Its important
to get that level of commitment early on. If not,
its better to look for another project or nd an-
other sponsor.
The best way to get a commitment is to
define the project around a problem of clear
interest to someone else. The manager of a
waste disposal facility isnt interested in re-
ducing the number of trucks he has. More
likely, hes willing to fight to keep them. Find-
ing out how to help the facility manager solve
a problem hes dealing with say, having
fewer angry customers through better dis-
patching practices is the best way to get
project commitment (assuming, of course,
that the facility managers ideas align with
corporate goals).
5IAkIING 8IG
While starting small can be effective, some-
times its neither possible nor desirable to do
so. If a big opportunity presents itself, and mo-
mentum within the organization is to move for-
ward, its almost always best to grab the reins
and go.
A good example outside the analytics do-
main is the introduction of human resource
planning software into an organization. The
software is designed to manage human re-
sources throughout an entire organization.
Even if it would potentially make sense to try
the software on a small part of an organiza-
tion before undertaking a corporate overhaul,
software sales agents sell a corporate vision,
and organizations purchase that same vision.
Such a large project will certainly be broken
into smaller, more manageable rollouts. But
the concept from the very beginning is, in a
word, big. A good example in the analytics do-
main is airline revenue management software,
where decisions on what seat inventory to sell
on one ight impact similar inventory decisions
on all other ights throughout the network.
In big projects, the focus shifts from convinc-
ing management that analytics is of value to
making the project successful. The very nature
of a big project is that the value decision has
been made. Nonetheless, as an analytics proj-
ect, part of the value will be how much money the
project makes or saves the organization based
on the decision support it provides. When hu-
man resource planning software is implement-
ed, the value in terms of improved workow
has already been accounted for. The same is
true of an airline revenue management system,
but airlines will nonetheless want after-the-fact
demonstrations of the value. How much money
did the revenue management system make last
month? Is it giving results that make sense?
Measurement and evaluation are common to
/ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
CkAW|, WA|K Ok kUN
BY E. ANDkEW BOYD
Wbclbcr you`rc louncbing on onolylics corccr or pro|ccl, is il bcsl lo
slorl big, smoll or somcwbcrc in lbc middlc?
P k O| l T C E NT E k
all analytics projects, and are a topic well ad-
dress in a future column.
5IAkI IN IH MI00L
There are, of course, intermediate proj-
ects. A good example is a grocery store chain
testing a new pricing strategy at an individual
store. Pilot projects of this nature require the
involvement of multiple individuals. The orga-
nization has enough faith in the effort to give
analytics a try, but the ultimate decision has
yet to be reached. In many ways, pilot projects
are the most difcult, since they carry the bur-
den of both developing corporate commitment,
like small projects, and making a project with
many moving parts a success.
Pilot projects seem like a good idea, and they
can be. An analytics project that demonstrates
value in the laboratory is shaken out in the
real world. Where problems arise is that the
real world brings analytics professionals with
new ideas face to face with professionals whove
been doing their job for many years. As a result,
pilot projects are often more fractious than help-
ful. With no clear go/no-go decision made by
management, theres endless room for bickering.
Theoretically, bringing together many different
opinions is a good idea, with the best decision
rising to the top. In practice, the go/no-go deci-
sion is a complex mix of facts and politics.
Some of the problems can be mitigated if
upper management conveys that a decision to
move forward has been made and that the pi-
lot project is an effort to iron out details. As a
result, its worth the effort to seek this level of
commitment before the project starts. If more
work needs to be done to demonstrate that the
underlying analytics has sufcient value, its
easier to do so before beginning a pilot project.
That doesnt mean ignoring those individuals
in the eld, outside the laboratory.
Quite the contrary. Analytics can only be
successful if it deals with all the real world
complexities of a problem. But its far more
efcient to air the facts, make a decision and
move forward than to be in a state of constant
indecision. Too often, the decision to undertake
a pilot project simply pushes problems into the
future that are better dealt with now.
5UMMAkY
Crawling, walking and running all have their
advantages and disadvantages when starting
down the path of analytics. But even if one ap-
proach was clearly best, the approach thats
ultimately taken is largely shaped by the cir-
cumstances. Common to all approaches and
really to all managerial decision-making is
the need to understand the environment, have
clear goals and choose a carefully reasoned
path a path to success.
Andrew Boyd served as executive and chief scientist at
an analytics firm for many years. He can be reached at
e.a.boyd@earthlink.net.
0 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TREOkETlCA||Y, BklNGlNG
TOGETREk NANY
Dl||EkENT OPlNlONS lS
A GOOD lDEA, WlTR TRE
BEST DEClSlON klSlNG TO
TRE TOP. lN PkACTlCE, TRE
GO/NO-GO DEClSlON
lS A CONP|EX NlX O|
|ACTS AND PO|lTlCS.
P k O| l T C E NT E k
One of the most influential mathematical
finance models ever is the so-called Gauss-
ian Copula model. Now consigned to the
dustbin of failed experiments, the Gaussian
Copula enjoyed untold stardom in the early
years of this century, becoming central in
the valuation and rating of the convoluted
mortgage-related assets that unleashed the
credit crisis.
The formerly glorious concoction prom-
ised to accurately estimate the joint default
probabilities and default correlations of a
bunch of debt securities pooled together.
It was widely embraced around the finan-
cial industry as a trusted indicator of the
likelihood that a lot of the (often dodgy)
mortgages making up those infamous Col-
lateralized Debt Obligations would turn
nasty concurrently. Due to its technical in-
tricacies, the Gaussian Copula typically de-
livered bland default correlation estimates,
facilitating the confidence-building AAA-ing
of junky subprime investments. As is now
only too obvious, the model failed drasti-
cally as numerous clustered defaults took
place rather more habitually than theoreti-
cally predicted.
The models failure would not seem
that surprising to those aware of the origi-
nal rationale for its embedding in Finance-
land. David Li, the father of the hitherto
reputable mathematical baby, initially got
his inspiration from academics engaged
in modeling the broken heart effect (cal-
culating the odds of a person dying once
their significant others have passed away,
an important concept in actuarial science).
Believing that default is like the death of
a company, so we should model this the
same way we model human life, Li saw it
only fit to apply the mathematics of death
correlation to the tackling of default corre-
lation. That is, the quantitative device that
aided and abetted the lethal trading in toxic
mortgage stuff first made its entry into the
markets because someone saw nothing
wrong with equating financial activity with
the workings of the human body. Thats as
odd as concluding upon seeing stock pric-
es go up and down that the science behind
elevator-building should be used to fore-
cast the S&P 500.
As would be plain to a 10-year old, just
because companies, like people, can die,
that doesnt imply that, barring the complete
loss of our common sense, we can analyze
debt defaults as the loss of a human life. No
wonder the Gaussian Copula suffered an
unmitigated defeat out there in the heart-
less markets.
Unfortunately, David Lis is not the only
instance of wildly inappropriate analogies
when it comes to financial modeling. Math-
ematicians and theoreticians appear inca-
pable of not attempting to equate market
activity to all sorts of odd, insultingly un-
related fields. For decades weve heard fi-
nancial economists and assorted scientists
tell us that market pros are like molecules,
and that if you know physics then youve
got the markets mapped. The ignored ram-
blings of hopeless cranks? Far from it.
Models borrowing from exactly such odd-
ness have been furiously applauded, em-
braced and granted the Nobel. Meanwhile,
markets and real traders kept busy trotting
along in a notoriously non-molecular fash-
ion, naturally.
These days, biological finance appears
to be the new in phrase. Bank portfolios
should be modeled on savannahs (appar-
ently tropical rainforests are to be avoided
at all costs). Regulators should heed the dic-
tums of marine experts. In epidemiologists
apparently lies the solution to too-big-to-
fail. Mathematical biologists, mirroring the
now-prevalent physicists, are reportedly try-
ing to jump into the financial bandwagon, at-
tempting to succeed where so many other
brainiacs fell by the wayside before. Upon
contemplating such process, the rest of us
should worry.
Bad things have afflicted the world fol-
lowing the adoption of the creations of those
bent on treating finance as something else.
1 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TRE PEkl|S O|
EOUATlNG
BANKEkS TO BEES
BY PAB|O TklANA
NATRENATlClANS
AND TREOkETlClANS
APPEAk lNCAPAB|E
O| NOT ATTENPTlNG
TO EOUATE NAkKET
ACTlVlTY TO A|| SOkTS
O| ODD, lNSU|TlNG|Y
UNkE|ATED |lE|DS.
V l E W P Ol NT
The crash of October 1987 was caused by
the use of a theory that espouses that stock
prices behave like dust particles. The re-
cent mayhem owes a lot of its misery not
just to mathematically confusing corporate
bankruptcy with human mortality, but also
to assuming market activity to be a realm of
chaos-devoid calmness. We should aim to
prevent the next big troubles from emerg-
ing via the adoption of a formula that de-
picts hedge fund managers as pollinating
bees.
A big reason why we know that the odd
mathematical homologies are unsound is
that the process would certainly not work
backward. Would a financier (even if quanti-
tatively skillful) be able to accurately map the
physical terrain, based solely on his market
career? Be able to model death? Be able to
model the survival rate of an ecosystem? Of
course not. For the simple reason that finan-
cial practice has little to do with the science
of physics, mortality or biology. Just like a
career trader would not be considered an
authority on fisheries management or atomic
studies, should non-financiers bearing irre-
mediably weird theorems be given credence
in Financeland?
Pablo Triana (climacover@gmail.com) is the author of Lecturing
Birds on Flying: Can Mathematical Theories Destroy The
Financial Markets?
2 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TRE BlkDS AND TRE
BEES: |ECTUklNG BlkDS
ON ||YlNG AUTROk
PAB|O TklANA WAkNS
|lNANClA| NODE|EkS
AGAlNST DEPlCTlNG
REDGE |UND NANAGEkS
AS PO||lNATlNG BEES.
V l E W P Ol NT
kEADEkSRl P SUkVEY
Give us your opinion:
http://vista-survey.com/survey/v2/survey2.dsb?ID=4850246745
HUMANIIAkIAN k5AkCH
GkOUP kACI5 IO HAIII
In the aftermath of the dev-
astating earthquake in Haiti,
one of the aftershocks facing
the country is not seismic in na-
ture, but concerns the struggle to coordinate
relief efforts in order to reach those in need
most effectively. INSEADs Humanitarian Re-
search Group (HRG), part of the INSEAD So-
cial Innovation Centre, has worked closely with
many humanitarian organizations such as the
International Federation of the Red Cross/ Red
Crescent, World Food Programme, World Vi-
sion International and Mdecins San Frontires.
By listening to the needs and experiences of
these partner organizations, INSEAD HRG has
obtained rst-hand knowledge of the issues
affecting humanitarian logistics in responding
to disasters. HRG has developed a rigorous
methodology to analyze these issues, develop
solutions and disseminate knowledge to those
working in and with the humanitarian sector.
We are closely monitoring events in Haiti
and have been in contact with many of our
partners who are working there, says Profes-
sor Luk Van Wassenhove (photo), director of
the Social Innovation Centre and founder of
INSEAD HRG. It is already a country with a
lot of poverty and weak infrastructure, so the
logistical challenges faced by humanitarian or-
ganizations on the ground will be great.
The tragedy in Haiti has prompted a rapid
international reaction and outpouring of aid.
The magnitude of the disaster combined with
severe infrastructure, security and communi-
cation problems illustrates the crucial need to
continue working on this topic.
For more on the INSEAD Social Innovation
Centre, its reaction to the Haiti Earthquake and
a video interview with Luk Van Wassenhove
and Rolando Tomasini, manager of INSEAD
HRG, click here: http://www.insead.edu/facul-
tyresearch/centres/isic/HaitiEarthquake.cfm
kAPI0 0LIVkY OF
VACCIN5, ANII0OI5
The ongoing threat of the
H1N1 u, future pandemics
and biowarfare have given new
urgency to pioneering research
done by Professor Eva Lee (photo) of Georgia
Tech on the rapid design and deployment of lo-
gistical strategies to deliver both vaccines and
antidotes during medical crises. Lee, in collab-
oration with the Centers for Disease Control,
is the creator of RealOpt, a decision support
system enabling real-time strategic and opera-
tional planning for large-scale systems model-
ing and optimization for public health.
RealOpt is designed to take the guess-
work out of mass dispensing of medical
countermeasures by providing good esti-
mates of resource needs and operational
performance. In numerous anthrax exercis-
es and flu vaccination events conducted in
the United States, emergency planners were
able to determine best facility layout, opti-
mal staffing and throughput, cost-effective
operations, and perform strategic and oper-
ational planning. The system provides rapid
scenario analysis and makes dynamic deci-
sion analysis instant and scalable.
A more mature system since rst deployed in
2003, RealOpt is used not only by public health
directors and coordinators (more than 1,600), but
also by some re departments, school districts
and the U.S. military for the purpose of setting
up vaccine dispensing sites. It has been used for
efcient clinic layout design, optimal resource al-
location and disease propagation analysis. Plan-
ning has been carried out for anthrax response
exercises, actual u vaccination clinic operations
and for hepatitis vaccination.
Lees work has become an integral part of
White House and DHS planning and discus-
sions for preventing and preempting health
care threats. Last year, the Department of
Health and Human Services invited Lee to the
White House for a public meeting to discuss
H1N1 vaccine distribution issues and to share
her knowledge on vaccination distribution, dis-
pensing and tracking strategies with state, lo-
cal and tribal public health directors.
Professor Lee also works with White House
Biodefense Policy and Medical Prepared-
ness Policy Directors on matters related to
*) u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
DlSASTEk kE|lE|,
BlODE|ENSE PO|lCY,
NANAGENENT lDEAS
lT lS A|kEADY A
COUNTkY WlTR A |OT
O| POVEkTY AND WEAK
lN|kASTkUCTUkE, SO TRE
|OGlSTlCA| CRA||ENGES
|ACED BY RUNANlTAklAN
OkGANlZATlONS ON TRE
GkOUND Wl|| BE GkEAT.
NE W S N A K E k S
emergency response, mass casualty mitiga-
tion and medical preparedness and policy.
RealOpt-regional is an interactive online
software enterprise that features visualization
tools and large-scale optimization. The system
equips users with spatial understanding of im-
portant landmarks in the region, assesses the
population densities and demographic makeup
of the region, and identies the most cost-ef-
cient network of dispensing sites and modali-
ties for effective population protection.
Besides specifying the number of necessary
healthcare workers, security personnel and oper-
ations personnel, the system displays the demo-
graphic mixture of the region to help emergency
planners identify appropriate personnel for spe-
cial needs (e.g., pediatric assistants, translators).
Economic analysis can be performed to identify
combinations of dispensing modalities (walk-
through, drive-through, public, private, mobile, or
postal) to accommodate the affected population
in a timely and effective manner.
Lees effort aligns particularly well with the
recent news that the Department of Health
and Human Services will review in rst-quarter
2010 how the nation can more quickly devel-
op and produce medical countermeasures for
public health emergencies.
'ANALYIIC5' MAK5
H8k'5 LI5I OF IOP I0A5
Competing on Analytics
wasnt just a best-selling book
by co-authors Tom Davenport
(photo) and Jeanne Harris
(photo), it was also one of 12
most inuential management
ideas of the millennium (so far)
according to a handful of edi-
tors at the Harvard Business
Review. Decades of invest-
ment in systems capturing transactions and
feedback nally yielded a toolkit for turning all
that data into intelligence, wrote the editors.
Operations research types, long consigned to
engineering realms like manufacturing sched-
uling, got involved in marketing decisions.
Managers started learning from experiments
that were worthy of the name.
Davenport and Harris both recently contrib-
uted articles to Analytics magazine. Davenport
wrote The Rise of Strategic Analytics for the
Executive Edge column in the Fall 2009 issue
(http://viewer.zmags.com/publication/2d674a
63#/2d674a63/5), while Harris was co-author
of a feature article, How to Organize Your
Analytical Talent, that appeared in the Janu-
ary/February 2010 issue (http://viewer.zmags.
com/publication/ed2f8ceb#/ed2f8ceb/16).
Along with Competing on Analytics, HBRs
list of most inuential ideas included: Share-
holder Value as a Strategy, IT as a Util-
ity, The Customer Chorus, Enterprise Risk
Management, The Creative Organization,
Open Source, Going Private, Behavioral
Economics, High Potentials, Reverse Inno-
vation and Sustainability.
For the complete story, click here (http://
blogs.hbr.org/hbr/hbreditors/2010/01/the_
decade_in_management_ideas.html).
** u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
ECONONlC ANA|YSlS
CAN BE PEk|OkNED TO
lDENTl|Y CONBlNATlONS
O| DlSPENSlNG NODA|lTlES
lWA|KTRkOUGR, DklVE-
TRkOUGR, PUB|lC, PklVATE,
NOBl|E, Ok POSTA|
TO ACCONNODATE
TRE A||ECTED POPU|ATlON
lN A TlNE|Y AND
E||ECTlVE NANNEk.
NE W S N A K E k S
<HGG><M>=%M><A&L:OOR<HGLNF>KL<:GLB@GB?B<:GMER
EHP>KK>O>GN>L'AHPLAHNE=<HFI:GB>LK>LIHG=8
IF WE DONT invent an answer
to this were history...
These words could serve
as a cautionary note to any managers who
ignore the importance of analytics in their
industry. They are attributed to Donald
Burr, the entrepreneur who helped found
People Express Airlines in 1981 and,
over the next four years, led it through
exceptional growth to revenues nearing
$1 billion reportedly the fastest growth
of a U.S. company in history to that time
[Cross, 1997]. The company worked with
a lean workforce and minimal overhead
while providing no-frills service. This
allowed it to offer airfares signicantly
lower than those of any major competi-
tor. By 1985, People Express was the fth
largest U.S. passenger carrier and was
directly challenging major U.S. airlines in
many of their key markets.
What Burr needed to invent an an-
swer to was an extraordinary action tak-
en by American Airlines. In January 1985,
American introduced deep discount fares
across its ight network, including every
route own by People Express. American
was a full-service airline with much higher
xed costs than those of People Express.
How could it sustain these prices without
threatening its own survival?
Americans secret was its new yield
management system, which permitted it
to maintain or increase its protability while
lowering fares for signicant amounts of
seat inventory on nearly every ight. This
was not short-term predatory pricing at a
loss the airline could sustain these pric-
es until competitors either responded by
kEVENUE NANAGENENT
lN TRE EkA O|
SOClA| NETWOkKlNG
BY YUkl |EVlN l|E|T AND 1E|| NcGl|| lklGRT
IF II WE
to thi
Th
ti
I
CAU T l ONA kY TA | E
*+ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
*, u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
increasing their service levels (and prices)
or went out of business. People Express did
attempt to change its business model but
ultimately failed, and the airline was forced
out of business two years later.
This is a classic example of what has
been called strategic analytics [Davenport,
2009] detailed, data-driven analyses sup-
porting outcomes of strategic importance
for a company. In the case of American
Airlines, the analysis involved careful cal-
culation of the number of discount seats on
each ight that could be sold early without
excessive revenue loss from late-booking,
high-fare customers. These calculations
depended critically on the probabilities or
odds of future levels of passenger de-
mand for full and discount fares based on
computer records from thousands of past
ights. In effect, Dallas-based American
played a winning game of Texas Hold-em
against an ill-equipped competitor.
American subsequently reported reve-
nue gains of $1.4 billion over three years
directly attributable to its yield manage-
ment system [Smith, Leimkuhler and Dar-
row, 1992]. Such publicized successes in
the airline industry were soon followed by
expansions into related businesses in the
travel and hospitality industries and, later,
into areas like broadcasting, event ticket
sales and seasonal goods retailing. Yield
management is now more commonly
called revenue management (RM) or its
newer counterpart, dynamic pricing.
The growth of the Internet and other
information technologies has had obvious
and important impacts on the spread of RM
for producers. There are, however, inter-
esting implications for RM in the spread of
Internet applications for consumers. Man-
agers of existing RM systems face genuine
challenges from consumer-based systems
in the age of social networking. A variety
of responses are possible, but the key is to
recognize the need.
What is revenue management? The
usual answer is something along these lines:
Charge the right price to the right custom-
er at the right time; however, this sounds
more like a slogan than an answer. Many
managers would rightfully ask: Whats new
about that? Indeed, this matching of price,
customer and time is the outcome of good
ol-fashioned haggling if all goes well, the
customer receives something for less than
they would have been willing to pay, and the
seller receives more than their minimum ac-
ceptable price. An economist would say that
the customer receives a consumer surplus
and the seller a producer surplus. A skilled
negotiator on the producers side can maxi-
mize his or her surplus by minimizing each
customers surplus (while leaving customers
somewhat satised). This, of course, means
that prices are likely to be different for every
consumer.
Revenue management aims to maxi-
mize a producers surplus in modern mar-
kets, where prices cannot be determined
one customer at a time. Instead, different
prices are charged to different segments
of customers according to their estimat-
ed willingness to pay. Negotiation is re-
placed with fencing that aims to limit
switching of customers from high-priced
to discounted products. In most settings,
economic surplus translates directly to
revenues, and, in industries like the air-
lines with high xed and low marginal
costs, revenues mean prots.
Firms practicing RM are not the only
winners differential pricing can extend
the possibility of purchase to a much wider
population of consumers who were formerly
priced-out of the market. In effect, there is
a partial transfer of consumer surplus from
well-healed consumers to those less fortu-
nate a form of free-market socialism!
Conditions appropriate for RM are
typically based on the airline and hotel
situations: perishable assets, nite ca-
pacity, low marginal costs and ability to
use differential pricing. Recent expan-
sions into areas such as retail sales have
blurred the distinction between RM and
other pricing practices, but regardless of
the terms used, the basic objectives re-
main the same. For further details of RM
theory, many excellent references are
available, including [Phillips, 2005] or
[Talluri and van Ryzin, 2004].
How are RM and dynamic pricing re-
lated? In its early days RM was an inventory
control technique. Prices for different mar-
ket segments were announced by market-
ing departments, and operational decisions
were limited to deciding how many assets to
allocate to each price class. As the strategic
importance of RM grew, the need for coor-
dination of prices and the quantities became
apparent. In modern systems, price uc-
tuations can occur either because different
predetermined price classes are opened or
closed (inventory control) or because prices
are being adjusted over time within market
segments (dynamic pricing). The choice to
use inventory control, dynamic pricing or both
gives analysts signicant exibility. The re-
sult looks the same to the customer prices
that uctuate over time but the differences
can be substantial in implementation.
Where is RM practiced today? RM
developments have been reported in
k . N . & S OC l A | NE T WOk K l NG
RE|P PkONOTE :G:ERMB <L
Its fast and its easy! Visit:
http://analytics.informs.org/button.html
*- u : G: ER M B <L F:@: S B G> ' <HF
all of the following areas: air, rail and
cruise travel; lodging, hotels and ca-
sinos; car rentals; sporting and enter-
tainment event ticket sales; retail sales,
particularly seasonal or short-life-cy-
cle goods; Internet search (click and
ad pricing); broadcasting (advertising
time); Internet service providers; cel-
lular telephone services; cable televi-
sion; manufacturing; energy markets;
sea and air cargo and freight; nancial
services; and health services.
Given the scope of this expansion, it
might be more sensible to ask where it
has not been applied. Good overviews
for most of these can be found in the
general RM references listed earlier.
Social networking and Strategic
Consumers what todays RM manag-
ers need to gure out an answer to.
Consumers can become aware of
the patterns of price reductions prac-
ticed by producers, particularly if these
are systematized by RM practices,
and become strategic; for example, de-
laying their purchases to times of an-
ticipated lower price. There is nothing
really new here; some cost-conscious
purchasers can always be counted on
to wait to the end of a model run for
a lower price. However, the capabilities
of the Internet open up the possibility
of entirely new levels of sophistication
for consumers. We classify these op-
portunities under the general heading
social networking, but potential devel-
opments certainly extend beyond the
range of Facebook and Twitter. Here
are some examples:
Buyers cooperatives are an old
phenomenon with new potential.
In their traditional form, such co-
ops aggregate purchase requests
for individual consumers or small
retailers and then use increased
buyer power to seek quantity
discounts or other favorable terms.
It is now feasible for consumers
who are informally linked on the
Web to quickly form a buyers club
and exercise similar power. Such
co-ops could form and dissolve in a
matter of days for the sole purpose
of acquiring the latest hot product
at a discount, or they could become
a permanent and expanding
reality. RM systems that use bulk
purchasing as a fence for their
market segments beware!
Online markets such as E-Bay
or Craigslist are widely used for
exchange of both old and new
assets. They are also used for
exchange of tickets for entertainment
or sporting events perhaps an
k . N . & S OC l A | NE T WOk K l NG
*. u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
innocent transfer for someone who
cant use the tickets, but more often,
high-tech scalping. It is well known
that holders of season passes to
popular venues can more than pay for
their passes in the secondary market.
Essentially, social networking can
create an informal futures market for
virtually any good or service that can
be transferred. (Airlines are protected
from this because airline reservations
are linked to purchasers and, with
heightened security, particularly difcult
to transfer.) These consumer behaviors
can be likened to tunneling under the
fences that suppliers have created to
segment their markets; for example,
a late-purchaser who gets a ticket
for something close to an advance
purchase price.
Managers of venues and ticket brokers
are well aware of these practices and
have been trying to devise strategies
to combat them, but it is not yet clear
what success they are having. (Efforts
of suppliers to enter the unofcial
markets themselves have led to
howls of unethical conduct [see, for
example, MusicFiX, 2009].) The key
conclusion is that any proposed or
existing RM system that involves
transferable assets needs to be alert
to the possibility of an unofcial futures
market in their business. All that is
needed for such developments are
prices that vary over time in a manner
that is somewhat predictable.
Transparency of pricing is perhaps
the most immediate impact of the
internet and social networking. Firms
accustomed to practicing differential
pricing selectively and opaquely
may now have their prices exposed
either ofcially through their own
Web sites or unofcially, through
consumer networking. This can have
the unintended effect of making some
consumers aware of more appealing
market segments for themselves.
Online pricing advisors amplify price
transparency. Companies such as
Expedia, Travelocity and Priceline
currently give consumers free passive
information about prices across the
travel and hospitality industries. We say
passive because such sites mainly
summarize current prices, or in the
case of Priceline, offer the possibility of
bidding for preferred prices. However,
there are also companies such as
Farecast (acquired by Microsoft in
2008) that boast: Unlike other travel
companies, Farecast predicts when a
user should buy a ticket based upon 175
billion points of previous airfare data.
Its engine can currently predict whether
airfare goes up or down up to a week
into the future with a claimed success
rate of 70%-75%. While Farecast has
a lot of competition, they claim it is the
only company which can predict future
prices [Crunchbase, 2010].
Evidently, it is possible to predict future
prices with a reasonable degree of accu-
racy by simply studying the outputs of RM
systems in the form of their prices over time.
How much longer will it be until a pricing
advisor moves to the next level: develop-
ing counter-RM analytical tools to model
a producers RM policies and improve con-
sumers purchasing strategies even more?
A company with such capabilities could
charge a subscription or other fee and be a
commercially viable enterprise doing noth-
ing but undoing RM systems!
A variant of this is exemplied by Bid-
LessTravel.com (BidLessTravel), a free
Web service that simply shares users infor-
mation about successful bids on Priceline.
This is perhaps the best current example of
the potential of social networking applica-
tions to increase consumer surplus.
PAkIIAL AN5Wk5?
STRATEGIC CONSUMERS and consum-
er choice modeling. One answer to the
complicated consumer reactions involved
in RM decisions is to explicitly model the
k . N . & S OC l A | NE T WOk K l NG
Purchaser-link reservations and security issues help protect airlines from informal futures markets.
*/ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
reactions of consumers to system actions.
The area of consumer choice modeling is
very active and likely to grow. For example,
discrete-choice models developed in the
1960s to model travelers choices in urban
transport have been generalized for much
more complex settings. No individual con-
sumer can be pinned-down by a model,
but average behaviors can be captured,
and the law of large numbers can take over
after that. Such models, integrated with RM
control systems, are being used effectively,
and these applications will likely increase
rapidly as both computer power and data-
capture technologies expand.
Can such modeling actually encompass
strategic consumer behavior? Our own re-
search on this question is consistent with
intuition; strategic consumers can signi-
cantly lower the revenues obtainable with
an RM system, in some cases below lev-
els that can be achieved with no RM at
all. However, companies that adjust for
consumer responses to their RM systems
can regain signicant amounts of the lost
revenue. In some cases, they can do bet-
ter than that; for example, the adroit use
of price matching guarantees can actually
increase revenues beyond those achieved
without strategic consumers [Aviv, Levin
and Nediak, 2010]. In effect, strategic con-
sumers can be treated like a new market
segment that can be accommodated, with
positive outcomes on both sides.
5PIkAL-0OWN FFCI5
IT WOULD SEEM that we have laid out
enough concerns to satisfy the most avid
worrier, but there is one more issue that
deserves mention. An inuential paper
[Cooper, Homem-de-Mello and Kleywegt,
2006] details the implications of an RM
system that imperfectly models consumer
reactions to RM policies. Under the right
circumstances, such a systems pricing or
inventory decisions can cause subsequent
shifts in consumer demand that lead to pol-
icies that reinforce those shifts, leading to
further policy changes and so on. The end
result can be seriously degraded policies.
When consumers responses are being
guided by counter-RM systems, is there a
possibility that such situations can be ex-
ploited by those systems? This needs not
be a deliberate reaction designed into the
counter-RM system; it could simply result
from that systems quick and consistent
reactions to the changes it detects on the
other side. A slow spiral-down could be
converted to a death-spiral! At the very least,
this reinforces the need for appropriate
modeling of consumer reactions. Also, the
use of intelligent reality checks in any sys-
tem should catch such system failures be-
fore they progress too far.
CONCLU5ION
REVENUE MANAGEMENT IS widely recognized
as a major success story for analytics. Our
cautionary notes here are not meant to sug-
gest that RM systems will soon be a thing of
the past. To the contrary, we expect RM ap-
plications to continue thriving and expand-
ing as more organizations reach the levels
of awareness and analytical capability they
need. By properly adjusting for increasing
capabilities on the consumer side, organiza-
tions can still achieve signicant gains. They
may have to give back some of that pro-
ducer surplus, but not so much that RM will
fail to be worth the trouble. However, Donald
Burrs fateful warning could now be sounded
for managers of existing RM systems who
ignore the importance of consumer reac-
tions in the age of social networking.
Yuri Levin (ylevin@business.queensu.ca) is an associate
professor and Distinguished Faculty Fellow in operations
management at Queens School of Business in Kingston,
Ontario, Canada. He has published widely in the general
areas of revenue management and dynamic pricing and
currently serves as an associate editor of Management
Science. Levin was guest editor (with Jeff McGill) of a
2009 special issue of the European Journal of Operational
Research on revenue management and dynamic pricing,
has consulted widely on consumer choice models and
optimization of pricing, and serves on advisory boards of
several companies.

Jeff McGill (mcgillj@queensu.ca) is a professor of
management science and Distinguished Faculty Professor
of Operations Management at Queens School of Busines.
McGill has researched revenue management and related
areas for more than 20 years and has eight years of
full-time industrial experience in product development
and operations research. He is former associate editor at
Transportation Science, associate editor at Operations
Research and senior editor at POMS Journal. McGill was
the 2008 winner of the INFORMS Revenue Management
and Pricing Section historical prize.
k . N . & S OC l A | NE T WOk K l NG
The authors thank Georgia Perakis of MIT who
suggested grouping strategic customers under the
general heading of social networking; Robert Phillips
of Columbia University and Nomis Solutions Inc.; and
Andy Boyd, former chief scientist at PROS Inc.
ACKNOWL EDGMENT
1. Aviv, Y., Levin, Y., and Nediak, M., 2010 (working
paper), Can Price Matching Mitigate Strategic
Consumer Behavior?
2. BidLessTravel (n.d.), retrieved from http://www.
bidlesstravel.com/.
3. Cooper, W., Homem-de-Mello, T., and Kleywegt, A.,
2006, Models of the spiral-down effect in revenue
management, Operations Research, pp. 968-987.
4. Cross, R. G., 1997, Revenue Management: Hard
Core Tactics for Market Domination, New York:
Broadway Books.
5. Crunchbase.com, 2010, Farecast (February 2010)
retried from www.crunchbase.com/company/farecast.
6. Davenport, T. H., 2009, The Rise of Strategic
Analytics, Analytics (fall 2009), www.
analyticsmagazine.com.
7. MusicFiX, 2009, The Boss angry over Ticketmaster
scandal (Feb. 5), retrieved Feb. 7 from NINEMSN:
http://music.ninemsn.com/article.aspx?id=741242.
8. Phillips, R., 2005, Pricing and Revenue
Optimization, Stanford, CA: Stanford University Press.
9. Smith, B. C., Leimkuhler, J. F., and Darrow, R. M.,
1992, Yield Management at American Airlines,
Interfaces, Vol. 22, No. 1, pp. 8-31.
10. Talluri, K., and van Ryzin, G., 2004, The Theory
and Practice of Revenue Management, Norwell, MA:
Kluwer Academic Publishers.
WORKS CI T ED
SUBSCkl BE TO :G:ERMB <L
Its fast, its easy and its FREE!
Just visit: http://analytics.informs.org/
*0 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
|Ok SYSTENS ANA|YSTS, GETTlNG O|| TO A GOOD STAkT ON ANY PkO1ECT lS CkUClA|.
A POlNT-BY-POlNT GUlDE |Ok |AUNCR.
THROUGHOUT OUR CAREERS, we
have found ourselves work-
ing from certain principles
that have served us well. We
share these managerial guidelines to pro-
vide the systems analyst with an appropri-
ate and hopefully helpful grounding. We
concentrate on the start of a project for we
have found that making a good start goes
a long way toward having a successful
nish. The principles are presented here
in outline form so the analyst can consid-
er them point-by-point and refer to them
easily as the need arises.
1. To nish strongly, start strongly.
Take the following actions when starting
a new project:
a. Introduce the team that will be
working toward the solution of
the problem. Give a very short
summary of each team members
qualifications, experience and the
role that each will play. You can
say something humorous about
some or all of the members. But,
dont make that the focus of your
introduction. Its not necessary to
discuss failures that individuals
have suffered.
b. Be aware of team member
limitations. No one can, nor
should, do it all. Instead, strive
for a representative balance of
responsibilities among the team
members depending on their
strengths and need for development.
This will require making choices and
changing what is not working.
c. Explain why the study is being
conducted. People may be quite
concerned that the study is being
done. Will it eliminate their position or
change their work function with which
SEVEN SUGGESTlONS |Ok
SUCCESS|U| SO|UTlON STAkTUPS
BY 1EkkY BANKS l|E|T AND KEN NUSSE|NAN lklGRT
THROU
have
ing fr
that ha
I
P k AC T l T l ONE k S & P k O1 E C T S
*1 u : G: ER M B <L F:@: S B G> ' <HF
they have become so comfortable?
There may be rumors about the
purpose of the study and what
its impacts might be. Dispel the
rumors; tell the truth, even if it might
have a negative impact.
d. Confirm the objective. This
complements the previous
guideline. Instead of dwelling on
what is not the intention of the
study, tell the customer what it is.
A complete success is unlikely
without a clear and common goal.
Write it down! The team needs
to be properly aligned from the
start; unless everyone knows
what is expected and is reminded
of it, they have little chance of
accomplishing it. People contribute
more if they understand what the
intention is.
2. Crisp at the beginning, crisp
at the end. Crispness is enhanced by
making sure that your objectives have
the following characteristics:
a. They need to be precise. For
example, dont say, Our objective
is to improve communication.
Thats too general of a concept.
Make the objective more specific,
such as, We want to decrease
customer complaints by 10
percent from the last quarter of
the current year to the first quarter
of next year.
b. They need to be achievable. If
you make the goal so vast that
it cant be achieved, it will be
virtually meaningless and will
undermine the spirit of the team
before they even start. Dont say,
We want to end lung cancer. We
certainly wish for that to happen,
but in many years and billions of
dollars spent, it hasnt happened
yet, and likely wont over the
course of your project.
c. They need to be understandable.
Upon reading the objective, it
should make sense to every team
member. If the objective is, We
want the Type II error to go below
8 percent, that may or may not
have any meaning to the team.
What is a Type II error? What does
8 percent represent? What is the
error rate now? By what date is
this to be accomplished?
d. They need to be measurable.
Avoid words that are subjective.
For example, We want to
have a substantial impact on
the percentage of defectives.
What is substantial? What is a
defective? For what product or
S U CC E S S | U | S TA k T U P S
*2 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
class of products? Over what time
period?
e. They need to be communicated
repeatedly. Team members can
easily forget the specifics of a goal
statement. Hence, repetition
is needed. Moreover, by keeping
the goal in the forefront, team
members work harder toward
its achievement.
3. Control expectations at the out-
set, exceed expectations at the nish.
This can be accomplished by taking the
following actions:
a. Temper expectations at the start, or
they may become outlandish at the
end. If expectations arent managed
early on, the customer might begin
to have many erroneous ideas about
what might be accomplished. These
ideas tend to grow as time passes.
b. Make sure that the customer
knows the boundaries of the
investigation, whats included and
whats excluded. The model should
be large enough to answer the
questions asked. It shouldnt be
any larger than that. For example,
in modeling the assembly of
automobiles, it was noticed that the
seats, among other components,
arrive just-in-time from an external
firm. Should the seat manufacture
be monitored? The answer is that it
depends on the questions asked.
c. Conrm the timeline. There are
two elements to an expectation
a what and a when. The best
solution is of limited or no value if
delivered after the time needed. Gain
the customers support by clearly
laying out the project plan. Armed
with this road map, the customer is
in the best position to support the
course of action and its timing. If the
timeline is not regularly conrmed, a
late completion is more likely. Team
members must be reminded of what
is on the front burner.
d. Dont agree too readily; put your
foot down. There is a tendency
among younger analysts to accept
every challenge. But, some requests
are completely out of line. Usually,
they are beyond the scope of work.
Or, they may be technically difcult.
Accept what is in the agreement,
or what might be an alternate. But,
add-on tasks each have a time
requirement, and when all of those
additions are made, there may be
surprises about what the analysts
have accepted as the amount of work
to be performed.
e. Continuously monitor and manage.
Midcourse corrections should be
anticipated. An analyst doesnt have
enough knowledge and foresight to
get it totally correct from the start.
Expect change; nothing runs as
smoothly as expected. This doesnt
mean to paralyze your progress
for fear of all that might go wrong.
Rather, it means to manage change.
Be quick to curb unrealistic ideas, yet
be open to value-add adjustments.
Learn to stay exible, for it is the best
antidote for change.
4. Manage the project by managing
the people. Consider the following:
a. Maintain control of the flow of
meetings by following an agenda.
An agenda is useful for several
reasons. First, if its on the agenda,
its more likely to be covered during
the meeting. Second, extraneous
topics are less likely to be raised if
they are not on the agenda. Third,
the most important topics can be
covered earliest so that the meeting
is less likely to end prior to their
discussion.
b. Respond to any questions that are
raised. Listen to each question and
respond factually. Dont sugarcoat
your response. The truth always
emerges. It is better if it comes from
the solution providers.
c. Moderate the customers concerns.
If there are concerns raised, discuss
them. Some may be legitimate, some
may not. Throughout the discussion,
inspire trust. This, in the end, will give
the customer the condence to act on
the recommendation.
e. Respect the customers time. Listen
attentively to the issues that are
raised, but move on once consensus
is reached. To the customer, how you
run meetings foreshadows how you
will run the project.
5. To get good answers from the
customer, ask good questions. Consid-
er these suggestions:
a. Prepare. By preparing your
questions in advance, you are less
likely to omit something important,
and your questions will tend to be
more logical and targeted. The
questions can be organized so that
each topic is fully explored before
moving to the next one.
b. Attempt to answer the questions
yourself. Can you comprehend how
S U CC E S S | U | S TA k T U P S
kEADEkSRl P SUkVEY
Give us your opinion:
http://vista-survey.com/survey/v2/
survey2.dsb?ID=4850246745
+) u : G: ER M B <L F:@: S B G> ' <HF
to logically answer the question?
Its not as important that you know
the correct answer as it is that the
question makes sense. If you were
asked, What is the distribution
of time between failures of a
machine, how would you answer
it? What denition of terms do you
need? Under what conditions is
the measurement made? Is the
distribution what is desired, or will
the clock times of failure and when
the machine is running again be
sufcient responses?
c. Ask lots of what if questions.
This shows the customer that you
are really thinking seriously about
their problem. What if the SKUs
are placed in the wrong bin? How
will the correction be made? What
activities occur when a machine
goes down? Who is responsible for
taking action?
d. Avoid asking questions that can
be answered by a yes or no
response. Dont ask if SKUs are
ever placed in the wrong bin.
Ask what causes this to occur.
What are the consequences of an
occurrence? You want the process
to reveal information, not preclude
it. Thus, be attentive to not just the
answer, but also your question.
e. Dont upset the customer by putting
them on the defensive.
i. Bad: Why did you store those
incoming pallets in the mixed
pallet makeup area rather than the
racks? Your tone is accusatory!
ii. Better: How do you decide where
to store incoming pallets? Here
you are asking for information,
which is a much better approach.
f. Ask the customer to predict the
solution. This is not an attempt to
get the nal answer as much as it is
to gain early insights and establish
value. The exercise actually
serves several purposes. First, it
determines whether the customer
has the same alternatives in mind
as you. Second, the customer
knows more about the system than
anyone else. You will denitely learn
something from the responses.
Finally, it establishes a baseline
of thought. In the end, should the
solution turn out differently, the
exercise legitimizes the projects
value. If the solution turns out the
same, the team needs to explore
why there was no knowledge gain.
6. Dont approach the assignment
with preconceived ideas. Follow these
suggestions:
S U CC E S S | U | S TA k T U P S
+* u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
a. Be a good systems analyst. Maintain
your skills. Read case studies. Read
the literature. Learn about new
software and its applications. Attend
technical conferences.
b. But, be a better listener. You
can learn a lot by listening to the
customer. Allow them to transcend
your thinking. They often hold the
secret to success. Its unlikely that
any individual will be able to explain
the entire system. So, you will have
to meet with many individuals. Take
adequate notes that will enable a
more complete understanding of the
system later.
c. Give the customer a chance to
explain. Dont lecture. This is not the
time to prove your worth. Listen to the
customer. Avoid putting words in the
customers mouth. You are looking
to draw out facts and encourage
participation. A successful solution
starts with a complete understanding
of the problem.
d. Solve the customers problem, not
yours. As the saying goes, If you
only have a hammer, you will see
every problem as a nail. So, you
need a comprehensive tool kit to
ensure that you work on the right
problem.
e. Dont leap to conclusions. Go
through the entire process of
analysis before deciding on the
solution. Dont make premature
judgments. State the problem,
determine what others have done
to solve this kind of problem, collect
data, analyze the problem, then,
select the best alternative and
implement it. Importantly, include
the customer in this process.
Selling the solution is done in
concert with the customer, not apart
from the customer.
f. Avoid criticizing the customers
response. If you violate this, the
customer is eventually going to stop
responding to you. And, the customer
may brand you as fault-nding so that
no other customer member responds
to you.
7. Lay out a good pathway for com-
munication. We suggest the following:
a. Get everything in writing. If it
becomes necessary to verify
some part of the understanding
concerning the assumptions,
solution procedure, data or
selected alternative, you had better
have it in writing. Oral records
are useless, and recall is prone to
error, advertently or inadvertently.
Writing everything down also
confirms to the customer that you
are listening.
b. Provide a project road map. The
customer needs to see the entire
project plan laid out in advance.
This helps the customer know when
resources will be needed and for what
purpose. With projects, people dont
like surprises.
c. Make the customer a part of the
solution team. If you do this, the
customer becomes your advocate. If
you dont, the customer can quickly
become an obstructionist.
d. Plan on having lots of interactions.
Have frequent meetings, many
milestones and early deliverables.
Results are sold all along the way, not
at the end.
e. Let the customer know. In the
spirit of good communication, if
a problem ever arises during the
project, inform the customer. Dont
withhold some negative aspect of
the solution for fear of the reaction.
The customer will be even angrier
if the issue is disclosed too late to
deal with it.
The guidelines given above have
worked well for us. And, they have
worked for a long time. They should
work for you as well, but perhaps there
are cultural differences in where you
operate and where we operate that
necessitate some refinement. There-
fore, our final bit of advice is to adjust
accordingly.
We might not have thought of all the
advice that could be given. If you can
think of something else, let us know,
and we will do our best to publicize it
widely.
Jerry Banks holds the title academic leader at
Tecnolgico de Monterrey in Monterrey, Mxico.
Previously, he retired from the faculty of the
School of Industrial and Systems Engineering
at Georgia Tech in Atlanta, Ga., and then he
was senior simulation technology advisor at
Brooks Automation. He may be contacted by
e-mail at jerry_banks@itesm.mx.

Ken Musselman is the strategic collaboration
director for the Regenstrief Center for Healthcare
Engineering at Purdue University. He also
currently serves as the immediate past-president
of the Institute of Industrial Engineers. For more
than 30 years, he has actively consulted in the
design, monitoring, planning and scheduling of
healthcare and manufacturing systems through
business activity monitoring and response,
simulation and advanced planning and scheduling.
He may be contacted by e-mail at kenmus@purdue.edu.
S U CC E S S | U | S TA k T U P S
1. Musselman, Ken, Guidelines for Success, in
Banks, J. (Ed.) Handbook of Simulation, John Wiley,
New York, 1998, pp. 721-743.
F URT HER READI NG SUBSCkl BE TO :G:ERMB <L
Its fast, its easy and its FREE!
Just visit: http://analytics.informs.org/
CkEATlVE lNTEGkATlON O| CONPUTEk, CONNUNlCATlON AND |OCATlON TECRNO|OGlES RE|P
A WlDE kANGE O| lNDUSTklES TRklVE lN Dl||lCU|T TlNES.
IN THE TWO YEARS since our last
survey of vehicle routing soft-
ware, the world has suffered
through a nancial collapse
unlike any since the 1920s. Sinking home
values, overseas wars, terrorism, threats of
global warming and turmoil in health care
have capped a decade that many would
prefer to forget. As we look forward, the
global nancial disaster has taught us one
thing: simply moving money from account
to account does not produce prosperity.
Instead, our well-being depends on building
economic activities that deliver goods and
services to real people, when and where
they need them.
Routing software companies are offer-
ing creative ways to integrate computer,
communication and location technologies
with algorithms and software, helping a
wide range of industries thrive in a period
of recovery. These technologies enable
routing software companies to provide so-
phisticated connectivity and productivity
tools to an increasingly mobile workforce
and widely distributed customers.
ICHNOLOGICAL FOUN0AIION5
A DECADE AGO, cell phones were for the
rich, map databases were still being de-
veloped and GPS-enabled navigation was
beyond the average consumer. Routing
software worked well, but it wasnt ex-
ible and relied on an imperfect platform,
making it difcult for the drivers to access
information on the road, and sometimes
creating a mismatch between the route
that drivers were directed to follow and
what they knew to be the best course.
According to Cyndi Brandt of UPS Lo-
gistics Technologies, new data sources
have recently become available, includ-
ing a more complete commercial roads
database, and true historical trafc data
based on real travel times. Chris Jones
of Descartes Systems Group also notes
an explosion of map data attributes and
capabilities. In the next year or two, we
ON TRE kOAD TO CONNECTlVlTY
BY 1ANlCE PAkTYKA l|E|T AND kANDO|PR RA|| lklGRT
IN THE
surve
ware,
throu
I
VE R l C | E k OU T l NG S O| T WA k E
++ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
+, u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
will have predictive travel speeds for
road segments down to 15-minute inter-
vals. Currently, drivers say the software
doesnt know everything that they know
about the routes. It is sometimes true
now, but wont be that way in the future,
Jones says.
Likewise, we are now seeing stronger
connectivity between routing softwares
traditional functions that of assigning
stops to drivers and placing them in an
optimal sequence with on-the-road nav-
igation. A computer-generated printout
listing turn-by-turn directions is becoming
more of an anachronism these days, when
so many drivers can receive voice com-
mands and dynamic map displays from
their phones or navigation devices. As
Jones notes, Customers and prospects
are really looking for combined solutions
that both route and navigate. The big
growth market is in the portable market,
he continues. You see eets that would
have never gotten GPS technology now
using it with a mobile phone.
According to Julian Stephens of MJC2,
Smart phones are changing the industry
in a big way because they allow real-time
data capture, which in turn enables real-
time re-optimization of the operation. This
is of particular interest to us because our
dynamic scheduling algorithms can make
use of these data to increase operational
efciency of mobile workforces. With this
type of on-the-road connectivity, MJC2
and other routing software companies
are investing in their real-time routing
features and algorithms.
Route Solutions Inc., for instance, has
integrated their routing solutions with por-
table TomTom devices. Estimated times
of arrival are generated automatically,
and dispatchers can remotely change
stops, directly through the device, while
drivers are in the eld.
5OFIWAk A5 A 5kVIC
ACCORDING TO GREG WIEITHOLTER of
Route Solutions, Desktop-based rout-
ing is going away. Instead, he says,
people want Web-based solutions, so
all parts of the organization can have vis-
ibility. This is now often accomplished
through the Software as a Service
(SaaS) model, whereby the software
vendor generates solutions and man-
ages data from their own servers. Alter-
natively, MJC2, for instance, has found
that many customers prefer to host the
software themselves as this gives them
more control and flexibility.
SaaS is changing the nancial model
of some software companies. As Jones of
Descartes notes, SaaS is the way the in-
dustry has headed. It means that custom-
ers pay as they go. It reduces the initial
payment of the customer and removes a
barrier to entry. UPS has a SaaS offering
aimed at smaller eets, which treat the soft-
ware costs like utility charges rather than
an up-front capitalization.
For the software companies themselves,
cash ows are changing from the initial
large balloon payment and lower mainte-
nance fees that come from purchase of li-
censes to more continuous income ows.
For some, the switch is a difcult nancial
transition that requires cash reserves. How-
ever, many routing companies have found
ongoing support nancially challenging un-
der the license model.
IH 8U5IN55 OF kOUIING
MANY OF THE ROUTING COMPANIES pre-
dict that consolidation in the industry is
VE R l C | E k OU T l NG S O| T WA k E
RE|P PkONOTE :G:ERMB <L
Its fast and its easy! Visit:
http://analytics.informs.org/button.html
+- u : G: ER M B <L F:@: S B G> ' <HF
likely to accelerate and that compa-
nies run by mathematicians will be
unsuccessful. Customers are looking
for stability and distribution system
expertise more than the latest algo-
rithms. Nevertheless, one of the chal-
lenges that all of these companies will
face is the ability to process data to
re-optimize routes in real-time. With
map data getting richer and real-time
traffic adding road segment data in
small time intervals, the processing
requirements will grow substantially.
The underlying architecture of these
offerings and how the processing is
distributed will be a critical factor in
how well each product fares.
XPAN5ION INIO NW IN0U5IkI5
ROUTING SOFTWARE IS being used in
an increasingly diverse set of indus-
tries, both in traditional truck pick-up
and delivery as well as in mobile ser-
vices, such as repair crews. Among
Descartes customers, Sleepys Mat-
tress Stores uses routing software
to schedule deliveries on the spot at
point-of-sale, within one-hour time
windows. At Appian Logistics Soft-
ware, it is being used to synchronize
trucks and plumbers so that icemak-
ers can be installed soon after a re-
frigerator is rolled in the door. Route
Solutions customers are using rout-
ing software to schedule multitudes
of merchandisers and installers, who
travel among retail outlets to install
product displays and signage.
One clear trend is that the com-
panies that use routing software in-
creasingly want to offer services that
set them apart from their competi-
tion, and these services demand both
faster solution times and improved
communication frequently generat-
ing new solutions on the spot. How-
ever, given the tough economy, cost
remains an important driver. While
desirable to route the same drivers
to the same customers each and ev-
ery day, that level of consistency can
be inefficient. As Hugh Gigante of
Appian notes, If we tell a customer
that it costs them $100 a day to keep
the same drivers servicing the same
customers, most fleets will decide it
isnt worth the cost. Instead, routes
can be continuously re-optimized so
that every vehicle is used to its max.
Defining features now are anything
that reduces windshield time, saves
VE R l C | E k OU T l NG S O| T WA k E
kEADEkSRl P SUkVEY
Give us your opinion:
http://vista-survey.com/survey/v2/
survey2.dsb?ID=4850246745
+. u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
gas and decreases wear and tear on
trucks, concludes Gigante. And trucks
need to be full.
8A5IC 5OFIWAk FAIUk5
THE ROUTING SOFTWARE surveyed pro-
vide a common set of basic capabilities:
geocoding addresses, i.e., locating the
latitude and longitude by matching the
address against data contained in a
digital map database;
determining the best paths through
street networks between pairs of
geocoded points;
solving vehicle routing problems,
entailing an assignment of stops to
routes and terminals, sequencing
stops and routing vehicles between
pairs of stops; and
displaying the results in both graphical
and tabular forms in such a way that
dispatchers can guide the solution
process and communicate results to
drivers, loaders and other personnel.
Applications differ somewhat accord-
ing to their target market, special features
and integration capabilities, as discussed
in the following section.
IHI5 YAk'5 5UkVY
SIXTEEN SOFTWARE VENDORS (12 North
American and four from Europe)
participated in this years routing software
survey, representing 22 products. Our
questionnaire was divided into sections
covering platform, algorithmic capabili-
ties, interfaces and features, applications,
system integration and background infor-
mation. All responses are self-reported
and unveried.
Platform. Windows remains the dom-
inant platform for routing software and
is available for nearly all software pack-
ages, with only one company offering a
Mac version of their software, and an-
other three in Linux and Unix. Fourteen
products are available in SaaS, two of
which are only available in SaaS. From
a hardware perspective, vendors gen-
erally recommend a high-end PC oper-
ating in the 3 GHz range, combined with
1 to 4 Gb of memory and up to 100 Gb
in hard disk space (much less if SaaS is
used). These figures have not changed
much in the last four years, again show-
ing that powerful routing software has
become easier to run on ones desktop
computer. In SaaS applications, com-
puting requirements are even smaller
than in the past.
Algorithmic capabilities. The algo-
rithms underlying routing products are gen-
erally proprietary, though typically involve
a combination of integer programming
methods and heuristics. QuantMethods
and Jeppesen were
more specic, re-
spectively stating
that linear program-
mi ng/branch-and-
bound and column
generation were
used. FreshStart Lo-
gistics, new to this
years survey, indi-
cated that their al-
gorithms are based
on articial intelli-
gence and expert
system approaches,
rather than tradi-
tional mathematical
programming. DNA
Evolutions response of construction and
improvement is likely representative of the
general class of heuristic methods used by
most routing vendors.
Vendors generally claim unlim-
ited problem size for their software,
but from a practical perspective, com-
putation time, memory size and disk
space bound product performance, so
it is important to test software on actual
problems. In this regard, most vendors
claim computation times in the one-to-
five-minute range for an average-sized
problem, described as the time to solve
a problem with 50 routes, 1,000 stops
and two-hour hard-time windows. (Keep
in mind that computation times are pro-
vided by the vendors and have not been
verified.) These times are similar to two
years ago.
Fast computation times are particu-
larly important in real-time applications,
such as when deliveries are scheduled
while the customer is on the phone or
when stops are inserted and scheduled
while vehicles are in the eld. Research-
ers in vehicle routing are well aware
that route-length approximations can be
very effective at producing good solu-
tions in a short amount of time. This year,
VE R l C | E k OU T l NG S O| T WA k E
Routing the same drivers to the same customers each day can be inefficient.
+/ u : G: ER M B <L F:@: S B G> ' <HF
vendors were tight lipped about use of
approximations.
Node routing is the capability to
assign and sequence discrete stops,
and arc routing is the capability to as-
sign and sequence street segments.
The former is needed most often, and
occurs when the driver visits 100 or
fewer locations per day. We believe it
is available on all products. Arc routing
is more specialized and occurs when
vehicles visit every (or most) address
on block segments, as in meter read-
ing, mail delivery and garbage pickup.
Most of the vendors claim they can
do both of these, along with real-time
routing, daily routing and route plan-
ning. However, a single routing pack-
age is unlikely to be adept at all of
these functions, and it is important to
select a vendor that has experience in
the planned application.
More than half of the products offer
some capability for real-time routing,
which could come in the form of real-
time re-routing of vehicles or real-time
stop scheduling. Six vendors (Appi-
an, Descartes, FreshStart Logistics,
MJC2, SAITECH and UPS Logistics)
have the ability to incorporate real-
time traffic, which is now more widely
available in major cities. This can en-
able a fleet to reschedule in response
to customer requirements, vehicle de-
lays or traffic conditions.
Most vendors claim the ability to
solve routing problems with soft time
windows. However, when asked for
specics, some simply indicate that
they represent a range of time or maxi-
mum allowed delay rather than a true
soft window approach. On the other
hand, IBM ILOG permits early and late
penalties, DNA Evolution windows are
fuzzy by default and MJC2 uses con-
gurable rules.
Interfaces and features. As a start-
ing point, basic features offered by most
include an ability to display routes and
stops on maps and edit these routes
with the drag-and-drop feature (i.e.,
click on a stop and move it to which-
ever route you desire). This enables
the dispatcher to modify the algorithm-
produced routes and is needed in prac-
tice to satisfy customer constraints. To
make these features work, products
need digital maps, such as the com-
mercial products from NAVTEQ (based
in the United States) and Tele Atlas
(based in Europe). Maps, which are not
VE R l C | E k OU T l NG S O| T WA k E
RE|P PkONOTE :G:ERMB <L
Its fast and its easy! Visit:
http://analytics.informs.org/button.html
+0 u : G: ER M B <L F:@: S B G> ' <HF
inexpensive, are often sold separately
and are chosen by customers accord-
ing to their requirements.
Integration. Real-time communica-
tion with drivers, as well as tracking
their locations, has become particu-
larly important, and most products
offer these features. This usually is
provided with vehicle-mounted de-
vices or mobile phones. Interfaces
with other software systems such
as order-entry and inventory man-
agement is also important for re-
tailers and distributors. We saw less
emphasis on this feature in this years
survey than in the past, though many
(Descartes, FreshStart Logistics, IBM
ILOG, MJC2, Prophesy, SAITECH) of-
fer integration with supply chain man-
agement. Other important features
include forecasts for delivery require-
ments, generation of load manifests
and load planning.
Applications. Whereas vendors
generally claim that their products are
designed to serve a broad range of
applications, most specialize in an in-
dustry sector. Specialization is largely
driven by interface requirements
both in terms of presenting informa-
tion in a manner that is useful to the
target user and in terms of interfacing
with business software systems and
hardware devices. Police, taxi and
emergency vehicle dispatch, for in-
stance, each demand special require-
ments that differ from the traditional
market of private fleets. They fall in the
realm of niche markets, even though
in theory they are just variations of
vehicle routing.
Vendors that are more experienced
in an industry will be better prepared
to consult on software installation and
more likely to have relevant features,
leading to a higher likelihood of suc-
cess. The optimization code might also
be different to account for the particular
network structure, for instance the hub-
and-spoke design of less-than-truck-
load (LTL) networks.
In our survey, most of the respon-
dents have specialized in private truck
eets, serving such markets as food
and beverage (e.g., Anheuser Busch,
Coca Cola and Gold Medal Bakery).
Jeppesens PlanOp is being used by
for-hire eets, such as Australia Post
and Purolator Courier.
Routing installations tend to require
a large degree of customization, as re-
ected in software prices, which often
runs in the tens of thousands of dollars.
As an alternative, SaaS provides rout-
ing services for a monthly fee, perhaps
as low as $99 per month. Beyond these
VE R l C | E k OU T l NG S O| T WA k E
+1 u : G: ER M B <L F:@: S B G> ' <HF
software costs, some level of consulting
is likely needed to ensure full integra-
tion into a eets information systems,
typically priced in the neighborhood of
$125 per hour. In terms of pure size,
many companies now claim more than
1,000 installations each.
General information. The accom-
panying directory provides contact in-
formation and product names for the
vendors. Pricing is available for some
vendors (in many cases, prices are
negotiable and depend on eet size).
Expect to pay $10,000 or more for the
software alone. Higher-priced products
generally offer more customized service,
a larger array of features and interface
capabilities, and specialized experience
in a particular industry. Price structures
do vary, so be sure to compare the full
installed cost before making a choice,
including license fees, installation and
maintenance costs, hardware and digi-
tal maps.
5LCIING A PkO0UCI
BEFORE PURCHASING A PRODUCT, eet
managers should rst assess the
degree to which routing solutions and
data must be integrated across multiple
sites and decide whether they would like
adopt the SaaS model or host their own
software. It also important to ask ques-
tions such as: How big of a problem will
be solved, measured in vehicles, stops
and terminals? How frequently will the
solution be updated, and how quickly
must the software generate a solution?
Can stops be scheduled on the spot, or
will they be generated in batches? Who
will use the software, how is the infor-
mation best presented to the user, and
are the users distributed among many
locations? Who will install and maintain
the software? With which software sys-
tems must the routing system interact?
Vendors should demonstrate that they
are experienced serving other eets
with similar requirements, and they
should provide references so that you
can verify claims.
Janice Partyka (jpartyka@jgpservices.net) is principal
of JGP Services, a marketing consulting practice that
focuses on mobile technologies and applications.
Randolph Hall (rwhall@usc.edu) is vice provost for
Research Advancement at the University of Southern
California.
VE R l C | E k OU T l NG S O| T WA k E
|l NK:
To view the directory of vehicle routing software products along with the survey
data, see: www.lionhrtpub.com/orms/surveys/vehicle_routing/vrss.html
Founded in 1907 as a messenger company
in Seattle, Wash., United Parcel Service (UPS)
has grown into a $51.5 billion corporation by
clearly focusing on the goal of enabling com-
merce around the globe. Today, UPS is a glob-
al company with one of the most recognized
and admired brands in the world. UPS has
become the worlds largest package delivery
company and a leading global provider of spe-
cialized transportation and logistics services.
Every day, UPS manages the ow of goods,
funds and information in more than 200 coun-
tries and territories worldwide.
Analytics, specically operations research
(O.R.), has been an important tool at UPS
for a very long time. In 1954, our then CEO,
George Smith, said the following: If we did not
have operational research, our rate of growth
might have been affected. As we grow in size,
our problems increase geometrically. Without
operational research, we would be analyzing
our problems intuitively only, and we would
miss many opportunities to get maximum ef-
ciency out of our operation.
In the 55 years that have elapsed since Smith
made this statement, UPS operations have be-
come signicantly more complex and distrib-
uted. In 1954, UPS ran a U.S.-only operation
that delivered and picked up non-committed
volume. In 2009, every UPS driver worldwide
services delivery volume with different commit-
ted times throughout the day. In 1954, our net-
work was limited to the United States and was
fed by over-the-road tractor/trailer movements or
movements on trains. In 2009, UPS worldwide
network includes an airline, built from scratch,
which includes 208 aircraft in service each day.
With 263 total aircraft in our inventory, UPS has
the ninth largest airline in the world.
UP5 5IkUCIUk
UPS small package operations are divided into:
1) Package delivery/pickup (the Brown
trucks you see in your neighborhood).
2) Hub sorting packages.
3) Feeder over-the-road package
transport, hub to hub or hub to delivery
center. Its called feeder because this
operation feeds packages to hubs and
delivery centers.
4) Airline package transport via our air
network.
UPS has developed and actively utilizes
operations research models and applications
in all four of these operating areas.
O.k. IN UP5 PACKAG
0LIVkY/PICKUP OPkAIION5
UPS continues to use O.R. to enhance
its award-winning Package Flow Technology
(PFT), a technology suite UPS utilizes to opti-
mize sortation, dispatch planning and delivery
in its small package operation. By leveraging
address and map data, customer data and
real-time GPS and by integrating O.R. algo-
rithms into PFT, UPS is testing O.R. algorithms
that will provide real-time guidance to drivers
in order to service customers in the most cost-
effective manner.
UPS is testing this new on-road integrated
optimization and navigation system in several
locations around the United States. The O.R.
algorithm balances optimality and consistency
to provide a solution that should drive cost ef-
ciencies while meeting service and customer
commitments.
O.k. IN UP5 HU8 OPkAIION5
UPS has three optimization tools to assist
with planning UPS hub operations. They can
be classied as long-term planning, medium-
term planning and short-term planning.
Hub and Feeder Network Optimization
(HFNO) is the long-range planning optimiza-
tion tool. HFNO is used to determine how to
best utilize our UPS capacity while maintain-
ing service commitments and minimizing trans-
portation costs.
HFNO is used for facility capacity needs
for at least 10 years into the future. HFNO is
also used to set strategy and provide policy
+2 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
ANA|YTlCS ol UPS
BY 1E|| WlNTEkS
AND DOUG NORk
An insidc look ol lbc lccbnicol lools cmpowcring o $50 billion dclivcry ond
logislics compony lbol opcrolcs in morc lbon 200 counlrics worldwidc.
C Ok P Ok AT E P k O| l | E
Headquar ters: Sandy Spri ngs, Ga.
Busi ness: The worl ds l argest package del i very company, UPS
del i vers more t han 15 mi l l i on packages a day t o 6. 1 mi l l i on cust omers
i n more t han 200 count ri es and t erri t ori es around t he worl d. UPS i s
al so a l eadi ng provi der of l ogi st i cs sol ut i ons.
Web si te: www. ups. com
COMPANY NAME: UNI TED PARCEL SERVI CE
,) u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
decisions. For example, HFNO has been used
to model what the UPS network would look like
prior to acquiring new companies. It gives UPS
analysts insight into how the added capacity
will t into the UPS network and provides ana-
lysts a high-level estimate on impact to operat-
ing cost.
Network Planners Toolkit (NPT) is the me-
dium-term optimization tool. NPT provides de-
cision support to assist planners in answering
the following questions:
What are service feasible alternate ows for
packages?
How can packages be re-routed to bypass
a sort or facility?
Can volume be taken out of one load to
create a direct load?
Can service be enhanced by owing
packages a different way?
Load Planning Assistant (LPA) is the short-
term planning tool. LPA is used to plan what
loads will need to be moved two weeks prior to
execution. The scope of LPA is to optimize a
single day for a single building up to four build-
ings simultaneously. This optimization is de-
ployed to all of our major hubs in the US.
LPA was developed in conjunction with
Castle Lab (Princeton University). UPS de-
veloped the interface and Castle Lab (Warren
Powell and Belgacem Bouzaiene-Ayari) devel-
oped the optimization. UPS also has a multi-
day, multi-building load planning optimization
running at Castle Lab. The next step will be to
move this multi-day, multi-building model out
of the lab and into UPS operations.
O.k. IN UP5 F0k OPkAIION5
Feeder Scheduling and Optimization Sys-
tem (FSOS) is an optimization system used by
UPS local feeder operations to generate fea-
sible, efcient feeder driver schedules which
meet all UPS and labor work rules, including
minimum and maximum driver hours and driver
domicile restrictions. FSOS starts by evaluat-
ing all individual feeder movements, and then
determines alternative ways for combining
these movements into individual driver sched-
ules. FSOS repeatedly improves the solution
until improvement stops. FSOS is typically run
quarterly in each of our operating districts.
In addition to the brown feeder eet, UPS
needs to hire third-party drivers and driver
teams (called Expediters in UPS lingo) to
help move the additional package volume the
company handles during the busy peak time-
frame leading up to the holiday season. The
planners use the Expedited Feeder Schedul-
ing (EFS) system to develop the routes these
drivers run. UPS planners develop a list of the
loads that will need to be moved by these third-
party carriers, and EFS strings these loads to-
gether into a system of round-trip routes that
moves all the loads within their specied time
window at minimal cost. The planners have es-
timated that it would take about 20-plus people
three to four months to develop a similar set of
routes using traditional manual processes.
O.k. IN UP5 AIkLIN OPkAIION5
While UPS created its airline in 1988, the
company has been using O.R. to help plan
and manage its air operation since UPS be-
gan acquiring its own aircraft in 1981. Over the
years O.R. tools have been used in a variety
of ways to support the air operation including
developing simulator training schedules for
UPS pilots to meet currency and training re-
quirements, helping determine how many air
containers need to be in position at all UPS
locations to support the peak operations time
between Thanksgiving and Christmas, and
developing operational plans for using leased,
small jet aircraft to ensure that packages still
meet their service commitments in the event of
delays in the UPS air system.
Currently, the airline is using the Volume,
Location, Aircraft Network Optimizer (VOLCA-
NO) to help plan the domestic U.S. next-day
operation. The initial VOLCANO model was
C Ok P Ok AT E P k O| l | E
With 263 total aircraft, UPS has the ninth largest airline in the world.
,* u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
lN|OkNS
|U|| PAGE AkTWOkK
G>P:KMPHKD
PAGE 31
developed in conjunction with Cindy Barnhart
at MIT. VOLCANO allows air planners to ex-
amine changing as much, or as little, of the
air network as is needed to solve aircraft and
air hub capacity problems. The system will de-
termine the least-cost air network needed to
move all packages given the available aircraft,
their operating restrictions, and constraints on
airport and air hub capacities. Recently the air
planners have been using VOLCANO to re-
duce the number of aircraft that UPS needs to
y to deliver all air packages, resulting in lower
costs to both UPS and its customers.
This model assumed that the package ow
was xed. Since the initial model, the O.R.
group has added the capability to change the
package ow while simultaneously generating
aircraft routes and schedules.
Operations Research will continue to be
an important factor in the success of UPS. As
the companys network and business become
more and more complex, UPS will rely on op-
erations research tools and techniques to as-
sist it in maintaining an efcient, reliable and
cost effective service.
Jeff Winters (jwinters@ups.com) manages a UPS Operations
Research group based in Timonium, Md., that specializes
in algorithm and application development for UPS small
package, hub and over-the-road operations and planning.
Doug Mohr (dmohr@ups.com) manages a UPS Operations
Research group based in Louisville, Ky., that specializes in
algorithm and application development for UPS hub and airline
operations and planning.
TRE SYSTEN DETEkNlNES
TRE |EAST-COST Alk
NETWOkK NEEDED
TO NOVE A|| PACKAGES
GlVEN TRE AVAl|AB|E
AlkCkA|T, TRElk OPEkATlNG
kESTklCTlONS, AND
CONSTkAlNTS ON AlkPOkT
AND RUB CAPAClTlES.
C Ok P Ok AT E P k O| l | E
Where do world-class organizations
go to showcase the strategic use
of O.R. and analytics?
High-Level Panel on
Supply Chain
and Logistics
Chip White, Professor &
Chair of Transportation &
Logistics, Georgia Tech
Chuck Holland, Vice
President of Process
Management, UPS
July Lemke, Executive Vice
President and CIO, Schneider
National
Stephen Lilly, Executive
Director, Distribution and
Logistics, Merck
Hear what top companies around the world have to say about what O.R. and analytics
can do for your organization.
Disney Parks and Resorts Mark Shafer, Senior Vice President of Revenue Management
Federal Energy Regulatory Commission Richard ONeill, Chief Economic Advisor
The Kroger Co. - John Osborne II, General Manager, IS Research and Development
MIT Sloan School of Mgmt. Center for Digital Business - Michael Schrage, Research Fellow
Genentech - Darren Johnson, Sr. Supply Chain Strategic Planner
Bucks New University UK - John Boylan, Professor, Head of Research-Faculty of Enterprise
and Innovation
Integral Analytics - Michael Ozog, Vice President & COO-Smart Grid Products
HP - Sanjay Singh, Vice President, Business Analytic Services
Kromite - Homie Razavi, Founder and Senior Consultant.
CDC - Leslie Lennert, MD, Healthcare Informatics Consultant and former Director, National
Center for Public Health Informatics, CDC
Stanford University - Warren Hausman, Professor of Management Science & Engineering
Intel - Erik Hertzler, Staff Engineer
Chevron - Frank Koch, Decision Analysis Practice Leader
Veritec Solutions - Warren Lieberman, President
Georgia Tech Ellis Johnson, Coca-Cola Chaired Professor, ISYE
Business Forecast Systems - Eric A. Stellwagen, Vice President
SAS Institute Michael Gilliland, Product Marketing Manager, Forecasting
Ford Research and Advanced Engineering - Erica Klamp, Technical Leader
Decision Strategies - Patrick Leach, Principal-Oil & Gas
Princeton - Hugo Simao, Senior O.R. Engineer & Professor
Schneider National - Ted Gifford, Distinguished Member of Technical Staff
Plus 40+ speakers
Thanks to our sponsor! http://meetings.informs.org/Practice2010
Plus
Edelman
Competition
and Gala
The best in
applied analytics
2 0 1 0 I NF OR MS C ONF E R E NC E ON O. R . P R AC T I C E
APPLYING SCIENCE
TO THE
Art of Business
April 18-20, 2010
Hilton Bonnet Creek
Orlando, Florida
Register NOW
Early Registration deadline
April 1, 2010
All meals for two days included!
SUBSCkl BE TO :G:ERMB <L
Its fast, its easy and its FREE! Just visit: http://analytics.informs.org/
Where do world-class organizations go to
showcase the strategic use of operations re-
search and analytics? For the past 10 years,
the venue of choice has been the INFORMS
(Institute for Operations Research and the
Management Sciences) conference on O.R.
practice, Applying Science to the Art of Busi-
ness. This years conference, scheduled for
April 18-20 at the new Hilton Bonnet Creek
Resort in Orlando, Fla., will once again demon-
strate the competitive advantage of analytics.
Executives and analysts from companies
such as Disney, HP, Chevron, Intel and Ford
will share best practices in applying analytics
to solve business problems. Academic speak-
ers from schools such as MIT, Princeton and
Stanford will provide training on strategy and
O.R. methodologies. Overall, the program will
offer high-quality, in-depth presentations by
more than 80 speakers.
Mark Shafer, senior vice president of
revenue management at Disney Parks and
Resorts, will open the conference with a ple-
nary talk on Game-Changing Analytics: A
Disney Perspective, exploring how one of
the most creative companies in the world in-
tegrates analytical innovation and decision-
making into its business. Richard ONeill,
chief economic advisor for the Federal Reg-
ulatory Commission, will deliver another
plenary on Better, Smarter Electricity Mar-
kets: How Optimization Software Can Save
Billions.
5UPk 8OWL OF APPLI0 O.k.
The Edelman Competition, known as the
Super Bowl of applied O.R., is always a high-
light of the conference. This year, six nalists
will showcase their high-impact analytics proj-
ects during the day on Monday (see sidebar),
with the 2010 winner announced at a gala
awards ceremony and banquet that evening.
(Admission to the banquet is included in the
conference registration fee.) The winner will
then repeat their presentation in a plenary ses-
sion on Tuesday.
Two other high-prole awards will be an-
nounced during the gala: the INFORMS Prize
and the Daniel H. Wagner Prize. The IN-
FORMS Prize honors a company that effec-
tively integrates analytics into organizational
decision-making and has repeatedly applied
O.R. principles in novel and lasting ways. The
Wagner Prize recognizes projects in which
strong mathematics has been applied to prac-
tical problems, emphasizing clear writing and
veriable practice success.
,+ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
WOk|D-C|ASS
ANA|YTlCS CON|EkENCE
BY TEkESA V. CkYAN
GANE-CRANGlNG
ANA|YTlCS: A DlSNEY
PEkSPECTlVE` P|ENAkY
KlCKS O|| NOkE TRAN
80 PkESENTATlONS.
Applying Scicncc lo lbc Arl ol Busincss` cvcnl
scl lor April 18-20 in Orlondo
P k AC T l C E P k E V l E W
(Preliminary speaker list)
Disney Parks and Resorts Mark Shafer, Senior Vice
President of Revenue Management
Federal Energy Regulatory Commission Richard ONeill,
Chief Economic Advisor
The Kroger Co. John Osborne II, General Manager, IS
Research and Development
MIT Sloan School of Management Center for Digital Business
Michael Schrage, Research Fellow
IBM Research Arun Hampapur, Director, Business
Analytics & Mathematical Sciences
Genentech- Darren Johnson, Senior Supply Chain Strategic
Planner
Areva T&D Avnaesh Jayantilal, Director, Marketing
Management Systems
Bucks New University UK John Boylan, Professor, Head of
Research-Faculty of Enterprise and Innovation
Integral Analytics, Inc. Michael Ozog, Vice President &
COO-Smart Grid Products
HP Sanjay Singh, Vice President, Business Analytic
Services
University of Cincinnati Jeffrey Camm, Professor, Chair of
Quantitative Analysis & Operations Management
Kromite, Inc. Homie Razavi, Founder and Senior Consultant
CDC Leslie Lennert, MD, Healthcare Informatics Consultant
and former Director, National Center for Public Health
Informatics, CDC
Stanford University Warren Hausman, Professor of
Management Science & Engineering
Intel Corp Erik Hertzler, Staff Engineer
Chevron Corp. Frank Koch, Decision Analysis Practice
Leader
Veritec Solutions Warren Lieberman, President
Georgia Institute of Technology Ellis Johnson, Coca-Cola
Chaired Professor, ISYE
Business Forecast Systems Eric A. Stellwagen, Vice
President
Forio Business Simulations Michael Bean, Senior
Consultant
SAS Institute Michael Gilliland, Product Marketing
Manager, Forecasting
Ford Research and Advanced Engineering Erica Klamp,
Technical Leader
Decision Strategies, Inc. Patrick Leach, Principal-Oil & Gas
Princeton Hugo Simao, Senior O.R. Engineer & Professor
Schneider National Ted Gifford, Distinguished Member of
Technical Staff
Innovative Decisions, Inc. Donald Buckshaw, Senior
Principal Analyst
Cypress Venture Group Bryan Guy, Managing Director
(Plus more than 40 other speakers)
REAL- WORLD COMMENTARY FROM
TODAY S BUSINESS LEADERS
CON|EkENCE WEBSl TE:
http://meetings.informs.org/Practice2010
5UPPLY CHAIN
AN0 XCUIIV FOkUM
Several new programs will be introduced this
year. A high-level panel discussion, organized
jointly by INFORMS and the Industry Studies
Association with funding from The Sloan Foun-
dation, is aimed at bringing cutting-edge re-
search and thinking on specic analytics topics
to practitioners. Chip White, professor and chair
of Transportation and Logistics at Georgia Tech,
has put together a panel of prominent execu-
tives to discuss trends and challenges in supply
chain management, transportation and logistics,
including Chuck Holland, vice president of Pro-
cess Management at UPS; Judy Lemke, CIO at
Schneider National; and Stephen Lilly, executive
director of Distribution and Logistics at Merck.
They will be joined by several academics known
for their research in supply chain management.
A new Executive Forum will reach out to mid-
and senior-level executives who want an intro-
duction to how O.R. can provide a competitive
advantage for their companies. Networking op-
portunities, a concierge service to identify talks
on the program of particular relevance and other
high-level takeaways will be offered.
IGHIY 555ION5
The conference program, incorporating 80+
sessions, has been designed by a 19-member
council of practitioners and academics, chaired
by William Klimack, senior consultant at Kromite,
Inc. Most of the talks will be delivered by invited
speakers, who were selected to address specic
topics by the council (see preliminary speaker list
sidebar). The invited program will be organized
along these tracks: sustainability, service science
and healthcare, decision analysis, managing un-
certainty and risk, methodology tutorials, supply
chain management and logistics, and software
tutorials. Additional Select Presentations were
identied through a rigorous review process in
which space constraints meant that less than half
the abstracts submitted could be selected.
For the rst time, the conference will also
feature a poster session. This brief presentation
format has attracted more than 45 submissions,
including descriptions of work that is still in prog-
ress or can only be presented at a high level.
The poster option lends itself to presentations
on projects that may not be extensive enough
for a 50-minute presentation but that offer useful
learning points for other O.R. practitioners.
CAkk-8UIL0ING PkOGkAM5
Two special programs within the confer-
ence are designed for future analytics leaders.
The Young Researcher Connection, for junior
faculty and young industry researchers, pro-
vides participants with insights into the criti-
cal problems facing business, helping them to
broaden their research agendas and providing
important networking contacts.
The INFORMS Professional Colloquium of-
fers a full-day, interactive workshop on real-world
,, u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TRE NEW Rl|TON
BONNET CkEEK kESOkT
Wl|| ROST lN|OkNS
CON|EkENCE AlNED
AT PkACTlTlONEkS.
P k AC T l C E P k E V l E W
Top executives will bring their perspectives and
advice on trends and challenges in this exciting
arena. They will be joined by leading academics
known for their work in supply chain management.
Moderator: Chip White, Professor & Chair of
Transportation & Logistics, Georgia Institute of
Technology
Panelists: Chuck Holland, Vice President of
Process Management, UPS; July Lemke, Executive
Vice President and CIO, Schneider National;
Stephen Lilly, Executive Director, Distribution and
Logistics, Merck
HIGH- LEVEL PANEL ON
SUPPLY CHAIN AND LOGISTICS
Where do world-class organizations go to
showcase the strategic use of operations re-
search and analytics? For the past 10 years,
the venue of choice has been the INFORMS
(Institute for Operations Research and the
Management Sciences) conference on O.R.
practice, Applying Science to the Art of Busi-
ness. This years conference, scheduled for
April 18-20 at the new Hilton Bonnet Creek
Resort in Orlando, Fla., will once again demon-
strate the competitive advantage of analytics.
Executives and analysts from companies
such as Disney, HP, Chevron, Intel and Ford
will share best practices in applying analytics
to solve business problems. Academic speak-
ers from schools such as MIT, Princeton and
Stanford will provide training on strategy and
O.R. methodologies. Overall, the program will
offer high-quality, in-depth presentations by
more than 80 speakers.
Mark W. Shafer, senior vice president of
revenue management and analytics, Walt
Disney Parks and Resorts, will open the
conference with a plenary talk on Game-
Changing Analytics: A Disney Perspective,
exploring how one of the most creative
companies in the world integrates analyti-
cal innovation and decision-making into its
business. Richard ONeill, chief economic
advisor for the Federal Regulatory Commis-
sion, will deliver another plenary on Better,
Smarter Electricity Markets: How Optimiza-
tion Software Can Save Billions.
794)6&3;03*%440-)(36
The Edelman Competition, known as the
Super Bowl of applied O.R., is always a high-
light of the conference. This year, six nalists
will showcase their high-impact analytics proj-
ects during the day on Monday (see sidebar),
with the 2010 winner announced at a gala
awards ceremony and banquet that evening.
(Admission to the banquet is included in the
conference registration fee.) The winner will
then repeat their presentation in a plenary ses-
sion on Tuesday.
Two other high-prole awards will be an-
nounced during the gala: the INFORMS Prize
and the Daniel H. Wagner Prize. The IN-
FORMS Prize honors a company that effec-
tively integrates analytics into organizational
decision-making and has repeatedly applied
O.R. principles in novel and lasting ways. The
Wagner Prize recognizes projects in which
strong mathematics has been applied to prac-
tical problems, emphasizing clear writing and
veriable practice success.
,+ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
WOk|D-C|ASS
ANA|YTlCS CON|EkENCE
BY TEkESA V. CkYAN
GANE-CRANGlNG
ANA|YTlCS: A DlSNEY
PEkSPECTlVE` P|ENAkY
KlCKS O|| NOkE TRAN
80 PkESENTATlONS.
Applying Scicncc lo lbc Arl ol Busincss` cvcnl
scl lor April 18-20 in Orlondo
P k AC T l C E P k E V l E W
(Preliminary speaker list)
Disney Parks and Resorts Mark W. Shafer, senior vice
president of revenue management and analytics, Walt
Disney Parks and Resorts
Federal Energy Regulatory Commission Richard ONeill,
Chief Economic Advisor
The Kroger Co. John Osborne II, General Manager, IS
Research and Development
MIT Sloan School of Management Center for Digital Business
Michael Schrage, Research Fellow
IBM Research Arun Hampapur, Director, Business
Analytics & Mathematical Sciences
Genentech- Darren Johnson, Senior Supply Chain Strategic
Planner
Aveva T&D Avnaesh Jayantilal, Director, Product
Management
Bucks New University UK John Boylan, Professor, Head of
Research-Faculty of Enterprise and Innovation
Integral Analytics, Inc. Michael Ozog, Vice President &
COO-Smart Grid Products
HP Sanjay Singh, Vice President, Business Analytic
Services
University of Cincinnati Jeffrey Camm, Professor, Chair of
Quantitative Analysis & Operations Management
Kromite, Inc. Homie Razavi, Founder and Senior Consultant
CDC Leslie Lennert, MD, Healthcare Informatics Consultant
and former Director, National Center for Public Health
Informatics, CDC
Intel Corp Erik Hertzler, Capital Supply Chain Staff
Engineer
Chevron Corp. Frank Koch, Decision Analysis Practice
Leader
Veritec Solutions Warren Lieberman, President
Georgia Institute of Technology Ellis Johnson, Coca-Cola
Chaired Professor, ISYE
Business Forecast Systems Eric A. Stellwagen, Vice
President
Forio Business Simulations Michael Bean, Senior
Consultant
SAS Institute Michael Gilliland, Product Marketing
Manager, Forecasting
Ford Research and Advanced Engineering Erica Klamp,
Technical Leader
Decision Strategies, Inc. Patrick Leach, Principal
Princeton Hugo Simao, Senior O.R. Engineer & Professor
Schneider National Ted Gifford, Distinguished Member of
Technical Staff
Innovative Decisions, Inc. Donald Buckshaw, Senior
Principal Analyst
Cypress Venture Group Bryan Guy, Managing Director
(Plus more than 40 other speakers)
REAL- WORLD COMMENTARY FROM
TODAY S BUSINESS LEADERS
CON|EkENCE WEBSl TE:
http://meetings.informs.org/Practice2010
With approximately 15,000 drivers and in-
dependent contractors, Green Bay, Wis.-based
Schneider National is one of the largest truck-
load carriers in North America. The companys
operations include about 14,000 tractors and
40,000 trailers, and Schneiders driver associ-
ates collectively cover more than ve million
loaded miles per day.
Needless to say, a transportation company
that big must deal with a host of complex lo-
gistic, strategic and tactical problems, not the
least of which is developing dispatching poli-
cies that help retain drivers in an industry his-
torically plagued by 100-percent annual driver
turnover rates while simultaneously meeting
the needs of customers and optimizing the de-
ployment of assets. For help in capturing the
dynamics of its massive eet in order to make
better corporate decisions regarding such is-
sues, Schneider turned to a team of internal
and external analysts. Among the questions
the team addressed:
What will be the impact of changes in
federal regulations governing drivers?
What is the best way to manage drivers
based in Canada?
Where should new drivers be hired?
How many teams (drivers that work in
pairs which can operate 20 hours each
day) should the company maintain?
Is it possible to get drivers home on a
regular basis and make commitments
of when a driver will be given time
at home?
The team included Ted Gifford, Jeff Day
and John Nienow of Schneider Nationals
Engineering and Research Department and
Hugo Simo, Abraham George and Warren
Powell of the Department of Operations Re-
search and Financial Engineering at Prince-
ton University. The team developed a model
known as the Tactical Planning Simulator
,. u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
L<AG>B=>KG:MBHG:E
KNGLPBMA:G:ERMB<L
MHIKH=N<>;>EB>O:;E>
K>LNEML%MA>FH=>EA:=
MH<EHL>ER?BM:<MN:E
?E>>MI>K?HKF:G<>%
F:M<ABG@MA>=><BLBHGL
H?:LDBEE>=@KHNIH?
=BLI:M<A>KLLNIIHKM>=
;RLM:M>&H?&MA>:KM
IE:GGBG@LRLM>FL'
Ik^l^gmZmbhgh_PZ`g^kIkbs^&pbggbg`phkd_hkZiieb^]ZgZerlblpbee[^
k^ikbl^]ZmBG?HKFLIkZ\mb\^<hg_^k^g\^'
:G:ER M B <L B G :<M B HG
With 15,000 drivers and 40,000 trailers, Schneider National trucks are a familiar sight on North American highways.
79440=',%-2
%2()<)'98-:)*3691
Several new programs will be introduced this
year. A high-level panel discussion, organized
jointly by INFORMS and the Industry Studies
Association with funding from The Sloan Foun-
dation, is aimed at bringing cutting-edge re-
search and thinking on specic analytics topics
to practitioners. Chip White, professor and chair
of Transportation and Logistics at Georgia Tech,
has put together a panel of prominent execu-
tives to discuss trends and challenges in supply
chain management, transportation and logistics,
including Chuck Holland, vice president of Pro-
cess Management at UPS; Judy Lemke, CIO at
Schneider National; and Stephen Lilly, executive
director of Distribution and Logistics at Merck.
They will be joined by several academics known
for their research in supply chain management.
A new Executive Forum will reach out to mid-
and senior-level executives who want an intro-
duction to how O.R. can provide a competitive
advantage for their companies. Networking op-
portunities, a concierge service to identify talks
on the program of particular relevance and other
high-level takeaways will be offered.
)-+,8=7)77-327
The conference program, incorporating 80+
sessions, has been designed by a 19-member
council of practitioners and academics, chaired
by William Klimack, senior consultant at Kromite,
Inc. Most of the talks will be delivered by invited
speakers, who were selected to address specic
topics by the council (see preliminary speaker list
sidebar). The invited program will be organized
along these tracks: sustainability, service science
and healthcare, decision analysis, managing un-
certainty and risk, methodology tutorials, supply
chain management and logistics, and software
tutorials. Additional Select Presentations were
identied through a rigorous review process in
which space constraints meant that less than half
the abstracts submitted could be selected.
For the rst time, the conference will also
feature a poster session. This brief presentation
format has attracted more than 45 submissions,
including descriptions of work that is still in prog-
ress or can only be presented at a high level.
The poster option lends itself to presentations
on projects that may not be extensive enough
for a 50-minute presentation but that offer useful
learning points for other O.R. practitioners.
'%6))6&9-0(-2+463+6%17
Two special programs within the confer-
ence are designed for future analytics leaders.
The Young Researcher Connection, for junior
faculty and young industry researchers, pro-
vides participants with insights into the criti-
cal problems facing business, helping them to
broaden their research agendas and providing
important networking contacts.
The INFORMS Professional Colloquium of-
fers a full-day, interactive workshop on real-world
,, u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TRE NEW Rl|TON
BONNET CkEEK kESOkT
Wl|| ROST lN|OkNS
CON|EkENCE AlNED
AT PkACTlTlONEkS.
P k AC T l C E P k E V l E W
Top executives will bring their perspectives and
advice on trends and challenges in this exciting
arena. They will be joined by leading academics
known for their work in supply chain management.
Moderator: Chip White, Professor & Chair of
Transportation & Logistics, Georgia Institute of
Technology
Panelists: Chuck Holland, Vice President of
Process Management, UPS; Judy Lemke,
Executive Vice President and CIO, Schneider
National; Stephen Lilly, Executive Director,
Distribution and Logistics, Merck
HIGH- LEVEL PANEL ON
SUPPLY CHAIN AND LOGISTICS
career guidance for masters and Ph.D. students
who are interested in practice careers. The col-
loquium will be held April 18, and students can
stay for the remainder of the conference.
Participants in both programs can register for
the full conference at a steep discounted rate of
$355; they must be nominated and selected to at-
tend. Limited nancial aid is available for the col-
loquium. The nomination deadline is March 30.
8)',2303+=;36/7,347
%2(73*87/-007 463+6%1
While the conference formally kicks off with
a Newcomers Orientation and Welcome Recep-
tion on April 18, the entire pre-conference day
is packed with valuable learning opportunities.
Attendees can take advantage of in-depth tech-
nology workshops presented by software solu-
tions companies, as well as a special workshop
on soft skills.
The Soft Skills Workshop provides a half-
day or full-day training in working successfully
with business decision-makers and the people
who implement decision analysis and O.R. so-
lutions. The morning session focuses on work-
ing with individuals and the afternoon session
focuses on groups. The sessions are designed
to complement one another, so attending the
full day will provide a comprehensive review.
)%60=6)+-786%8-326%8)
Early registration rates of $870 for INFORMS
members and $970 for nonmembers are avail-
able until April 1. Organizations can take advan-
tage of a team discount rate of $715 (by April
1) when they send three or more attendees. All
meal functions (breakfasts, lunches, breaks, the
Sunday reception and the Monday Edelman
banquet) are included in all registration fees.
The conference venue, the Hilton Orlando
Bonnet Creek Resort, is an upscale complex lo-
cated in a private natural setting on 482 acres and
surrounded on three sides by Walt Disney World
Resort. The reservation cut-off date is March 12.
For continuing updates, information on ho-
tel reservations and to register, go to http://
meetings.informs.org/Practice2010, phone
800-343-0062 or 401-722-2595 or visit meet-
ings@informs.org.
Teresa V. Cryan (Terry.Cryan@INFORMS.ORG) is the director of
meetings for INFORMS.
,- u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TRE SO|T SKl||S
WOkKSROP PkOVlDES
TkAlNlNG lN WOkKlNG
SUCCESS|U||Y WlTR BUSlNESS
DEClSlON-NAKEkS.
P k AC T l C E P k E V l E W
Six nalists will compete for the most prestigious
prize in applied O.R., the Franz Edelman Award.
The competition will take place April 19 and is
open to all conference registrants. That evening,
the six nalists will be honored and the 2010
winner announced at a gala ceremony and
banquet. The winners of two other important
practice prizes the INFORMS Prize and the
Daniel H. Wagner Prize will also be announced.
Edelman nalists include: Delaware River
Basin Commission (U.S.), Deutsche Post DHL
(Germany), INDEVAL (Mexico), New Brunswick
Dept. of Transportation (Canada), Procter &
Gamble (U.S.) and Sasol (South Africa).
EDELMAN COMPETITION:
THE BEST IN APPLIED ANALYTICS
Mark W. Shafer, senior
vice president of revenue
management and analytics,
Walt Disney Parks and Resorts
Presentation: Game-
Changing Analytics: A Disney
Perspective
Richard ONeill, Chief Economic
Advisor, Federal Energy
Regulatory Commission
Presentation: Better, Smarter
Electricity Markets: How Better
Optimization Software Can
Save Billions
THOUGHT- PROVOKING
KEYNOTE PRESENTATIONS
kEADEkSRl P SUkVEY
Give us your opinion:
http://vista-survey.com/survey/v2/survey2.dsb?ID=4850246745
career guidance for masters and Ph.D. students
who are interested in practice careers. The col-
loquium will be held April 18, and students can
stay for the remainder of the conference.
Participants in both programs can register for
the full conference at a steep discounted rate of
$355; they must be nominated and selected to at-
tend. Limited nancial aid is available for the col-
loquium. The nomination deadline is March 30.
ICHNOLOGY WOkK5HOP5
AN0 5OFI 5KILL5 PkOGkAM
While the conference formally kicks off with
a Newcomers Orientation and Welcome Recep-
tion on April 18, the entire pre-conference day
is packed with valuable learning opportunities.
Attendees can take advantage of in-depth tech-
nology workshops presented by software solu-
tions companies, as well as a special workshop
on soft skills.
The Soft Skills Workshop provides a half-
day or full-day training in working successfully
with business decision-makers and the people
who implement decision analysis and O.R. so-
lutions. The morning session focuses on work-
ing with individuals and the afternoon session
focuses on groups. The sessions are designed
to complement one another, so attending the
full day will provide a comprehensive review.
AkLY kGI5IkAIION kAI
Early registration rates of $870 for INFORMS
members and $970 for nonmembers are avail-
able until April 1. Organizations can take advan-
tage of a team discount rate of $715 (by April
1) when they send three or more attendees. All
meal functions (breakfasts, lunches, breaks, the
Sunday reception and the Monday Edelman
banquet) are included in all registration fees.
The conference venue, the Hilton Orlando
Bonnet Creek Resort, is an upscale complex lo-
cated in a private natural setting on 482 acres and
surrounded on three sides by Walt Disney World
Resort. The reservation cut-off date is March 12.
For continuing updates, information on ho-
tel reservations and to register, go to http://
meetings.informs.org/Practice2010, phone
800-343-0062 or 401-722-2595 or visit meet-
ings@informs.org.
Teresa V. Cryan (Terry.Cryan@INFORMS.ORG) is the director of
meetings for INFORMS.
,- u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TRE SO|T SKl||S
WOkKSROP PkOVlDES
TkAlNlNG lN WOkKlNG
SUCCESS|U||Y WlTR BUSlNESS
DEClSlON-NAKEkS.
P k AC T l C E P k E V l E W
Six nalists will compete for the most prestigious
prize in applied O.R., the Franz Edelman Award.
The competition will take place April 19 and is
open to all conference registrants. That evening,
the six nalists will be honored and the 2010
winner announced at a gala ceremony and
banquet. The winners of two other important
practice prizes the INFORMS Prize and the
Daniel H. Wagner Prize will also be announced.
Edelman nalists include: Delaware River
Basin Commission (U.S.), Deutsche Post DHL
(Germany), INDEVAL (Mexico), New Brunswick
Dept. of Transportation (Canada), Procter &
Gamble (U.S.) and Sasol (South Africa).
EDELMAN COMPETITION:
THE BEST IN APPLIED ANALYTICS
Mark Shafer, Senior Vice
President of Revenue
Management, Disney
Parks and Resorts
Presentation: Game-
Changing Analytics: A Disney
Perspective
Richard ONeill, Chief Economic
Advisor, Federal Energy
Regulatory Commission
Presentation: Better, Smarter
Electricity Markets: How Better
Optimization Software Can
Save Billions
THOUGHT- PROVOKING
KEYNOTE PRESENTATIONS
kEADEkSRl P SUkVEY
Give us your opinion:
http://vista-survey.com/survey/v2/survey2.dsb?ID=4850246745
,1 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*) ( +)*)
that you would be part of the people aiding and
abetting government expenditure on thermo-
nuclear war.
Nothing could be further from the truth!
In my many years in the mathematics divi-
sion I can honestly say that not one study done
by this department has ever been of any use to
anyone on the real side of the house. We have
been outrageously paid to do what we pleased.
And we pleased to do pure mathematics.
To get straight to the point, if you join us,
not only are you NOT accepting tainted mon-
ey, you are diverting government money that
was going to bomb design and improvement
to support your salary so you may continue
to campaign for peace. And remember, every
dollar you absorb is a dollar that didnt go into
the design and improvement of thermonuclear
devices. If you join us and work very hard you
can divert more money from atomic weapon
studies than any other way.
So you see, the best way you can siphon
away bomb-directed money and support peace
is to join us!
Yours for peace,
(Name withheld)
Gene Woolsey (robertwoolsey83@comcast.net) is professor
emeritus at the Colorado School of Mines, a past president of The
Institute of Management Sciences (TIMS) and a former editor of
several journals including The Production & Inventory Management
Journal, Interfaces, The Journal of Operations Management and
The Transactions of The Institute of Industrial Engineers.
BGFRF:GRR>:KL
BGMA>F:MA>F:MB<L
=BOBLBHGB<:GAHG>LMER
L:RMA:MGHMHG>
LMN=R=HG>;RMABL
=>I:KMF>GMA:L>O>K
;>>GH?:GRNL>MH
:GRHG>HGMA>K>:E
LB=>H?MA>AHNL>'
E :L M PHK =
@:FL
?NEEI:@>:KMPHKD
G>P:KMPHKD
I:@>,1
OPTIMIZATION
Europe
GAMS Software GmbH
Eupener Strasse 135-137
50933 Cologne, Germany
phone
+49-221-949-9170
fax
+49-221-949-9171
mail
info@gams.de
web
http://www.gams.de
USA
GAMS Development
Corporation
1217 Potomac Street, NW
Washington, DC 20007, USA
phone
+1-202-342-0180
fax
+1-202-342-0181
mail
sales@gams.com
web
http://www.gams.com
www.gams.comm
High-Level Modeling
The General Algebraic Modeling System (GAMS) is a
high-level modeling system for mathematical
programming problems. GAMS is tailored for complex,
large-scale modeling applications, and allows you to
build large maintainable models that can be adapted
quickly to new situations. Models are fully portable
from one computer platform to another.
Wide Range of Model Types
GAMS allows the formulation of models in
many different problem classes, including
Linear (LP) and Mixed Integer Linear (MIP)
Quadratic Programming (QCP) and
Mixed Integer QCP (MIQCP)
Nonlinear (NLP) and Mixed Integer NLP (MINLP)
Constrained Nonlinear Systems (CNS)
Mixed Complementary (MCP)
Programs with Equilibrium Constraints (MPEC)
Conic Programming Problems
Stochastic Linear Problems
GAMS Integrated Developer Environment for editing,
debugging, solving models, and viewing data.
Scheduling and Planning at BASF
Close cooperation between logistics, information services and the scientific computing group of
BASF, Prof. Dr. C. A. Floudas (Princeton University), Dr. A. V. Eremeev and Dr. P. A. Borisovski (Omsk
Branch of Sobolev Institute of Mathematics SB RAS), SAP AG, and Mathesis GmbH led to a number
of successfully deployed applications based on exact and hybrid optimization techniques. One of the
results is a novel modeling approach of batch and continuous plants:
State-task network formulation resulting in mixed-integer linear program
Unit-specific, event-specific continuous-time formulations
Hybrid methods and decomposition schemes to handle large instances
Tight lower bounds derived from auxiliary models
Implementation in GAMS with parallel GAMS/CPLEX
New interfacing technology and integration approaches to connect to SAP-APO
Used on a daily basis to improve planning and scheduling
State-of-the-Art Solvers
GAMS incorporates all major commercial
and academic state-of-the-art solution
technologies for a broad range of problem
types, including global nonlinear optimi-
zation solvers.
A>EI IKHFHM> :G:ERMB <L
Its fast and its easy! Visit: http://analytics.informs.org/button.html
With approximately 15,000 drivers and in-
dependent contractors, Green Bay, Wis.-based
Schneider National is one of the largest truck-
load carriers in North America. The companys
operations include about 14,000 tractors and
40,000 trailers, and Schneiders driver associ-
ates collectively cover more than ve million
loaded miles per day.
Needless to say, a transportation company
that big must deal with a host of complex lo-
gistic, strategic and tactical problems, not the
least of which is developing dispatching poli-
cies that help retain drivers in an industry his-
torically plagued by 100-percent annual driver
turnover rates while simultaneously meeting
the needs of customers and optimizing the de-
ployment of assets. For help in capturing the
dynamics of its massive eet in order to make
better corporate decisions regarding such is-
sues, Schneider turned to a team of internal
and external analysts. Among the questions
the team addressed:
What will be the impact of changes in
federal regulations governing drivers?
What is the best way to manage drivers
based in Canada?
Where should new drivers be hired?
How many teams (drivers that work in
pairs which can operate 20 hours each
day) should the company maintain?
Is it possible to get drivers home on a
regular basis and make commitments
of when a driver will be given time
at home?
The team included Ted Gifford, Jeff Day
and John Nienow of Schneider Nationals
Engineering and Research Department and
Hugo Simo, Abraham George and Warren
Powell of the Department of Operations Re-
search and Financial Engineering at Prince-
ton University. The team developed a model
known as the Tactical Planning Simulator
,. u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
SCRNElDEk NATlONA|
kUNS WlTR ANA|YTlCS
TO PkODUCE BE|lEVAB|E
kESU|TS, TRE NODE| RAD
TO C|OSE|Y |lT ACTUA|
||EET PEk|OkNANCE,
NATCRlNG TRE DEClSlONS
O| A SKl||ED GkOUP O|
DlSPATCREkS SUPPOkTED
BY STATE-O|-TRE AkT
P|ANNlNG SYSTENS.
Prcscnlolion ol Wogncr Prizc-winning work lor opplicd onolysis will bc
rcpriscd ol lN|OkNS Proclicc Conlcrcncc.
A NA |Y T l C S l N AC T l ON
With 15,000 drivers and 14,000 trailers, Schneider National trucks are a familiar sight on North American highways.
(TPS) that produces a highly detailed simu-
lation of fleet operations, capturing the dy-
namics of drivers and loads at a very high
level of detail.
As the authors note, To produce believ-
able results, the model had to closely t ac-
tual eet performance, matching the decisions
of a skilled group of dispatchers supported by
state-of-the art planning systems. To capture
the behavior of drivers in a realistic way, it was
necessary to model drivers using 15 sepa-
rate attributes. All work rules had to be repre-
sented to capture driver productivity. We also
had to model customer service requirements,
and other operational details such as driver
relays and the proper handling of geography-
constrained drivers [such as the Canadian and
regional drivers].
The probl em i s formul ated as a very
l arge-scal e stochasti c opti mi zati on prob-
l em to capture the col l ecti ve i ntel l i gence
of the di spatch center. When the team
model ed operati ons at the l evel of detai l
requi red both for proper cal i brati on and
to provi de sensi ti vi ty to key pol i cy stud-
i es, the resul t was a stochasti c l i near
programmi ng wi th an effecti vel y i nfi ni te
number of rows.
The problem was solved using approxi-
mate dynamic programming, which over-
comes the high dimensional state variables
using methods from machine learning. This
logic captured the critical ability of the dis-
patch center to anticipate the impact of de-
cisions made now on the future. This logic
made it possible to capture not only a 15-
dimensional attribute describing each driver,
but also uncertainties in loads (demands)
and travel times. Considerable attention was
put into capturing a host of real-world details
so that the model closely matched a number
of performance metrics.
The work not only saved Schneider Na-
tional millions of dollars in multiple applica-
tions since its introduction two years ago, it
also earned the 2009 Daniel H. Wagner Prize
for Excellence in Operations Research Prac-
tice from INFORMS (Institute for Operations
Research and the Management Sciences).
Gifford, Day, Nienow, Simo, George and
Powell describe their prize-winning work in
the paper, Approximate Dynamic Program-
ming Captures Fleet Operations for Sch-
neider National.
The winner was announced at the IN-
FORMS Annual Meeting last October following
a competition involving ve nalists. The other
nalist presentations included:
A Queuing Model-Based System for
Semiconductor Production Planning at IBM
Hub Group Implements a Suite of OR
Tools to Improve its Operations
Extending Bass for Improved New Product
Forecasting at Intel
Optimizing Helicopter Transport of Oil Rig
Crews at Petrobras
The award-winning Schneider National
presentation will be reprised at a special ses-
sion and honored at the awards gala dur-
ing the INFORMS spring practice meeting in
Orlando, Fla., April 18-20 (see page 32).
The Wagner Prize is given in memory of the
late Dr. Daniel H. Wagner. While president of
his own practice-oriented consulting rm, Dr.
Wagner brought many high-quality mathema-
ticians into the operations research commu-
nity, leading to signicant applications for U.S.
Navy, U.S. Coast Guard and several other
organizations. The prize honors Dr. Wagner
by emphasizing the quality and coherence of
analysis used in practice. Dr. Wagner strove
for strong mathematics applied to practical
problems, supported by clear and intelligible
writing. This prize recognizes those principles
by emphasizing good writing, strong analytical
content and veriable practice successes.
,/ u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
TRE WOkK NOT
ON|Y SAVED
SCRNElDEk NATlONA|
NU|Tl-Nl||lONS O|
DO||AkS lN NU|TlP|E
APP|lCATlONS, lT
A|SO EAkNED TRE
200 DANlE| R. WAGNEk
PklZE |Ok EXCE||ENCE
lN OPEkATlONS
kESEAkCR PkACTlCE.
WAG NE k P k l Z E W l NNE k
Wagner presentation videos:
www.livewebcast.net/INFORMS_
AM_Wagner_Prize_2009
Wagner Prize-winning paper:
www.castlelab.princeton.edu/Papers/Wagner_
Jan012010.pdf
Wagner Prize information:
www.informs.org/wagnerprize
INFORMS Spring 2010 Practice Conference:
http://meetings.informs.org/Practice2010
Schneider National:
http://viewer.zmags.com/publication/840980b8
#/840980b8/48
FOR MORE ON THE STORY
SUBSCkl BE TO :G:ERMB <L
Its fast, its easy and its FREE! Just visit: http://analytics.informs.org/
Some years ago there was a nameless Na-
tional Laboratory somewhere in this great land
of ours. The laboratorys main reason for exis-
tence was to improve the design of various kinds
of thermonuclear devices, build them and store
them. But like all government laboratories, its
reason for existence slid glacially and asymp-
totically toward pure research. The way to get
promoted or raised was to publish more papers
in the respectable journals in the eld. The
most far-out department was the mathematics
group. In theory their job was to provide what-
ever applied mathematics and optimization that
any other part of the plant might need. The truth,
however, was that its members happily had been
doing, were doing and proposed to continue to
do, pure research in whatever area of pure
mathematics ran their motor.
Now the mathematics department was an
aggressive recruiter and scouted out mainly
those university math departments that sat-
ised its requirement of outstanding ability in
pure mathematics. The result was, with a few
exceptions, the mathematics department was
made up of rather weird haystack-haired peo-
ple who adored pure mathematics, chess and
Kriegspiel. Kriegspiel took place every day at
noon and is, in essence, playing chess where
you and your opponent have a board and piec-
es but cannot see each others board. Only the
referee and spectators can do that. The ref-
eree tells you when it is OK to move a piece
to an unoccupied square and when you take
a piece. In short, it is simulated warfare pretty
close to reality. But I digress.
Many people who worked at this laboratory
were blissfully unaware that it was operated by a
phone company under contract with the Atomic
Energy Commission. What that meant in prac-
tice was as follows: If you applied for a math-
ematics position in parts of the phone company
and that organization didnt nd you desirable,
they would automatically forward your applica-
tion, as a courtesy, to the math department at
the National Laboratory where it would land on
the math department heads desk. In time a very
bright, young Ph.D. from a prestigious nameless
western school we shall call Berkeford, applied
to a mathematics division of the phone compa-
ny. He was brought out but, alas, didnt make the
cut. A week or so later he received a letter offer-
ing him an interview in the math department at
the National Laboratory.
This young man was a Class A Ban the
Bomber San Francisco pot-using hippie and
wrote a letter to the department head stating
how outraged he was to be contacted by a
bomb factory. The letter continued in an ex-
plosive way that he would see us all in Hades
before even considering working for such an
evil organization that designed thermonucle-
ar devices with which to kill people. He also
sent a copy of this letter to the vice president
of human resources at the phone company,
demanding that the process of forwarding
peaceful applicants to that den of sin and iniq-
uity should be stopped, forthwith!
Our department head handled this rather
well. He posted the young mans letter on the
bulletin board and challenged any member of
his department to write a reply; the author of the
best response would receive a bottle of Chteau
Latte-Rothschild. It wasnt long before some re-
ally witty letters were put up, including one that
everyone agreed was the winner. For political
reasons, it couldnt be sent as it was not classi-
ed, but it was too truthful. The letter follows:
Dear Dr. Hippie,
As I am sure you know, your letter telling us
all to go to hell provoked some interesting re-
actions here. There was a contest for the best
reply, which follows.
Firstly, you must realize that at one time this
facility was really pretty much a bomb factory, no
question. However, as the years passed, its con-
nection to anything real and anything having to
do with the design, improvement and storage of
thermonuclear devices has become a minuscule
part of our expenditure. Our math department
is really an ivory tower full of Ph.D.s. Our job is
to push back the frontiers of pure mathematics
while being paid a salary to research on what we
damned please while being compensated with
salaries that can only be described as amazing,
municent or wonderful.
You stated in your letter that you didnt want
to join us because this would make you a con-
tributor to the proliferation of thermonuclear
devices. And by accepting a salary you said
,0 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*)
A RlklNG lNClDENT
lN TRE DESEkT
BY GENE WOO|SEY
TRE NATRENATlCS
DEPAkTNENT WAS NADE
UP O| kATREk WElkD
RAYSTACK-RAlkED
PEOP|E WRO ADOkED
PUkE NATRENATlCS,
CRESS AND KklEGSPlE|.
| A S T WOk D
,1 u : G: ER M B <L F:@: S B G> ' <HF : G: ER M B <L u F:K <A ( : I K B E +)*) ( +)*)
that you would be part of the people aiding and
abetting government expenditure on thermo-
nuclear war.
Nothing could be further from the truth!
In my many years in the mathematics divi-
sion I can honestly say that not one study done
by this department has ever been of any use to
anyone on the real side of the house. We have
been outrageously paid to do what we pleased.
And we pleased to do pure mathematics.
To get straight to the point, if you join us,
not only are you NOT accepting tainted mon-
ey, you are diverting government money that
was going to bomb design and improvement
to support your salary so you may continue
to campaign for peace. And remember, every
dollar you absorb is a dollar that didnt go into
the design and improvement of thermonuclear
devices. If you join us and work very hard you
can divert more money from atomic weapon
studies than any other way.
So you see, the best way you can siphon
away bomb-directed money and support peace
is to join us!
Yours for peace,
(Name withheld)
Gene Woolsey (robertwoolsey83@comcast.net) is professor
emeritus at the Colorado School of Mines, a past president of The
Institute of Management Science (TIMS) and a former editor of
several journals including The Production & Inventory Management
Journal, Interfaces, The Journal of Operations Management and
The Transactions of The Institute of Industrial Engineers.
lN NY NANY YEAkS
lN TRE NATRENATlCS
DlVlSlON l CAN RONEST|Y
SAY TRAT NOT ONE
STUDY DONE BY TRlS
DEPAkTNENT RAS EVEk
BEEN O| ANY USE TO
ANYONE ON TRE kEA|
SlDE O| TRE ROUSE.
| A S T WOk D
GANS
|U|| PAGE AkTWOkK
G>P:KMPHKD
PAGE 38
OPTIMIZATION
Europe
GAMS Software GmbH
Eupener Strasse 135-137
50933 Cologne, Germany
phone
+49-221-949-9170
fax
+49-221-949-9171
mail
info@gams.de
web
http://www.gams.de
USA
GAMS Development
Corporation
1217 Potomac Street, NW
Washington, DC 20007, USA
phone
+1-202-342-0180
fax
+1-202-342-0181
mail
sales@gams.com
web
http://www.gams.com
www.gams.comm
High-Level Modeling
The General Algebraic Modeling System (GAMS) is a
high-level modeling system for mathematical
programming problems. GAMS is tailored for complex,
large-scale modeling applications, and allows you to
build large maintainable models that can be adapted
quickly to new situations. Models are fully portable
from one computer platform to another.
Wide Range of Model Types
GAMS allows the formulation of models in
many different problem classes, including
Linear (LP) and Mixed Integer Linear (MIP)
Quadratic Programming (QCP) and
Mixed Integer QCP (MIQCP)
Nonlinear (NLP) and Mixed Integer NLP (MINLP)
Constrained Nonlinear Systems (CNS)
Mixed Complementary (MCP)
Programs with Equilibrium Constraints (MPEC)
Conic Programming Problems
Stochastic Linear Problems
GAMS Integrated Developer Environment for editing,
debugging, solving models, and viewing data.
Scheduling and Planning at BASF
Close cooperation between logistics, information services and the scientific computing group of
BASF, Prof. Dr. C. A. Floudas (Princeton University), Dr. A. V. Eremeev and Dr. P. A. Borisovski (Omsk
Branch of Sobolev Institute of Mathematics SB RAS), SAP AG, and Mathesis GmbH led to a number
of successfully deployed applications based on exact and hybrid optimization techniques. One of the
results is a novel modeling approach of batch and continuous plants:
State-task network formulation resulting in mixed-integer linear program
Unit-specific, event-specific continuous-time formulations
Hybrid methods and decomposition schemes to handle large instances
Tight lower bounds derived from auxiliary models
Implementation in GAMS with parallel GAMS/CPLEX
New interfacing technology and integration approaches to connect to SAP-APO
Used on a daily basis to improve planning and scheduling
State-of-the-Art Solvers
GAMS incorporates all major commercial
and academic state-of-the-art solution
technologies for a broad range of problem
types, including global nonlinear optimi-
zation solvers.
RE|P PkONOTE :G:ERMB <L
Its fast and its easy! Visit: http://analytics.informs.org/button.html

You might also like