THE WEST COAST PORT SLOWDOWN and eventual longshoreman lockout last October
focused national attention on the delicate condition of America's supply
chain. An estimated $300 billion worth of goods flow through West Coast
ports every year, and as the giant container ships that carry it all sat
bobbing, anchored at sea and unable to off-load for 10 days, companies
across the country started to run out of inventory and assembly lines
began to grind to a halt.
Mitsubishi, out of engines and transmissions (which it imports from Japan),
suspended production of Eclipse convertibles and Galant sedans at its
plant in Normal, Ill. GM and Toyota ceased operations at a shared assembly
facility in Fremont, Calif., for one week until parts could be flown in.
Boeing's production of 767 and 777 airliners was disrupted because Asian-built
cargo doors and fuselage panels were delayed at sea.
Those and other interruptions were the product of a 20-year-old, 180-degree
about-face in the business world's approach to inventory management. Companies
used to measure their muscle by the size of their inventory. Bigger was
better. Vast warehouses filled to capacity ensured efficient assembly
lines and guaranteed that, come hell or high water, production would never
stop. Who cared about carrying costs? They would be erased by sales.
But now that equation has changed. Today, most inventory is considered
excess, and overstuffed warehouses represent missed opportunities and,
worst of all, wasted capital. Inventory has become such an anathema that
companies regularly risk shutting down their assembly lines in order to
hold less of it. Which is why the relatively brief port shutdown had such
a profound impact.
In the big picture, American business has succeeded in its quest to run
lean. U.S. Department of Commerce figures show that from 1981 to 2000,
inventory as a percentage of gross domestic product (GDP) fell 46 percent,
from 8.3 percent to 3.8 percent. During that same period, the total GDP
more than tripled (from $3.1 trillion to $9.9 trillion) while the total
amount tied up in inventory didn't even double, rising from $747 billion
to $1.48 trillion.
But almost all these gains in inventory reduction happened from 1981
to 1991, and the past 10 years have not seen much improvement. The inventory
carrying rate (a way of expressing the money spent to store inventory
as a percentage of its value), which was 34.7 percent in 1981, fell to
22.7 percent in 1992 but rose to 25.3 percent in 2000. And inventory as
a percent of GDP held steady at 3.8 percent from 1992 to 2000.
Playing Hot Potato with Inventory
But now, after 10 years of passing the inventory hot potato around, manufacturers
and suppliers are realizing that this is a game no one wins. If a manufacturer
attempts to reduce inventory by refusing to accept parts until a certain
point in its production cycle, it has to let its supplier knowin advancewhen
that moment will come. Otherwise, the supplier's warehouse costs will skyrocket.
And those costs will eventually be bounced back to the customer, the manufacturer.
Rather than being eliminated, inventory has been pushed down into the
lower reaches of the supply chainfrom manufacturers to top-tier
suppliers to lower-tier suppliers. GM, for example, improved inventory
turnsa common metric that measures total cost of goods sold divided
by average inventory, and serves as a valuable indication of how often
a company sells out its inventory (the higher the better)55.2 percent
between 1996 and 2001. However, the company that supplies its tires, Goodyear,
saw its inventory turns decline 21 percent during that same time. In other
words, GM left Goodyear holding the bag.
The lesson here is that when you get rid of inventory, you must replace
it with information.
Of course, that's easier said than done.
Up until now, companies have tended to confuse information with IT, but
after a decade spent throwing big moneyvia large ERP-type systemsat
the inventory problem, they are beginning to identify the information
they really need in order to make their supply chains work for everyone
concerned. CIOs now are focusing on smaller projects that target specific
information gaps in the supply chain. And new technologies are helping
them do just that. These projects generally cost in the tens of thousands
or low-hundreds of thousands as opposed to millions of dollars. They address
either the supply side, using information to sync with suppliers, or the
demand side, using information to right-size and right-time production.
Individually, they help drive out cost. Together, they can revolutionize
Big Rigs, Big Money, Big Problems
Cashman Equipment, a $250 million Las Vegas-based Caterpillar dealership,
sells and leases construction machinery ranging from the large (backhoes
and bulldozers) to the enormous (mining vehicles with 7-foot tires and
beds large enough to hold 32 pickup trucks). Recently, a drop in gold
and silver prices caused many Nevada-based mining companiesCashman's
customersto move their operations to South America where the ground
is softer and the cost of digging lower. The slumping economy also means
that rentals are up and sales are down.
"We need the right mix of machines in order to have the best margin,"
says CIO Bill Glassen. "We can't have machines languish in the yard."
Each piece of Cashman's large equipment costs between $250,000 and $1.5
million, and when the company can't sell it in a timely manner (a few
weeks to six months, depending on the product), it has to pay annual finance
charges, which can add up to $54,000 for every million dollar's worth
Glassen figured the best way to reduce inventory was to use all the sales
and trend information trapped in Cashman's database to get a more accurate
read on demand. At the time, the data was available only to people willing
to spend hours burrowing through ceiling-high stacks of computer printouts,
or as Glassen says, "The information was available but not accessible."
Rather than overhauling Cashman's systems and databases to throw IT at
the problem, Glassen spent about $12,000 in September on a demand-planning
system from Sunnyvale, Calif.-based Hyperion. The software has a specific
goal: Dig out the information buried deep in Cashman's databases and turn
it into reports on sales and rental trends. For example, the company recently
ran a report forecasting how many skid-steer loadersa one person
mini-loader used in landscapingit could expect to sell this year.
The report culled and analyzed, among other pieces of information, past
sales, the number of contractors in Nevada that use the machines and the
rate at which they are traditionally replaced.
Once the information became accessible and Cashman was able to predict
how many skid-steer loaders and other machines it could expect to sell,
the company could adjust its orders appropriately. In the first month
after the system became operational, Glassen says, Cashman reduced inventory
by $8 million without adversely affecting sales.
When Is a Sale Not a Sale? When Nobody Knows About It.
The best system in the world is worthless if no one uses it. A bad system
no one uses may be even worse. That was the situation at Palm, the Milpitas,
Calif.-based handheld computer maker. "When we spun away from 3Com
[in 2000], we were very successful very quickly," says Senior Vice
President of Global Operations Angel Mendez. "We went through hypergrowth.
Then [in 2001] we hit a very difficult period. Poor product introduction
and a stall in sales concurrent with a slowdown for PDAs left us with
a huge glut in inventory. We took a huge inventory write-down in the $240
Palm never saw it coming. Its forecasting system relied on regional sales
reps to dial in and enter sales reports and forecasts. Their information
wasn't inaccurate, per se, but it was usually late and often incomplete.
Salespeople found it too difficult to input information. That, Mendez
says, resulted in a 30 percent compliance rate. As a result, forecasts
were unreliable, and Palm had to carry excess inventory in order to guarantee
that it could meet orders.
In response, Palm added the demand-planner module to its existing SAP
system starting in April 2001. Although Palm considered integrating another
vendor's demand-planning software with its SAP ERP system, Mendez says
that because Palm already had a major investment in SAP, and since the
demand planner was easy for salespeople to use, it was the right investment.
All that was left to do was to make sure the sales force understood the
importance of complete and timely forecasts. Between June 2001 and November
2002, Palm reduced inventory from $240 million to $38 million. Inventory
turns jumped from three to 15 by the end of the first quarter.
Decisions in the Dark
Juniper Networks, an $887 million Sunnyvale, Calif.-based network equipment
maker, outsources the manufacture of all its routers to contractors that
ship the finished goods to Juniper's customers. So, says CIO Kim Perdikou,
it doesn't actually have any inventory. "But we have a responsibility
to share that burden" with the contract manufacturers, she says.
And that sharing affects the bottom line: If Juniper can help its contractors
lower their inventory charges, it stands to share in the savings.
Perdikou says that Juniper is philosophically opposed to big projects
where the scope of the software and the time it takes to integrate it
can distract from a project's business goal. So in 2001, when Juniper
wanted to get better inventory information from its suppliers rather than
attempt to integrate their ERP systems with its own Oracle system, Juniper
purchased and deployed a supply chain visibility system from Palo Alto,
Calif.-based Valdero. (Juniper won't say how much it paid, but Valdero
says that the system typically sells for between $300,000 to $500,000.)
Among other things, the software tracks inventory levels at partner companies,
either tapping EDI transmissions or through daily updates. Since Juniper
now has this information updated continuously, rather than in printed
monthly reports, it can make design decisions that take its partners'
inventory into account.
For example, Juniper could study 40 proposed engineering changes with
the goal of finding the five that would best reduce product cost. "We
may decide in engineering that we are going to change a component that
on paper makes [a product] more cost-effective," says Perdikou. "But
that isn't taking into account the effect on inventory at the contract
manufacturer." In other words, Juniper may want to replace a particular
part with a less expensive option, but if the contractor has a large stockpile
of the old part, then it stands to take a huge financial hitand
ultimately pass that cost on to Juniper.
"If you don't understand what the inventory is, you are really making
a lot of these decisions in the dark," says Perdikou.
A Disk in Time Saves Nine
Network Appliance, the $795 million Sunnyvale, Calif.-based enterprise
storage solu- tions provider, is currently in the middle of a year and
half long ERP project. Yet when it had to reduce its service inventorythe
parts it uses to replace broken devices in the fieldit stayed away
from big supply chain names like i2 Technologies and Ariba, and instead
cobbled together smaller, more targeted pieces of software.
"If you have big problems, you may need a big project," says
CIO Scot Klimke. "In this case, however, we had a problem that was
very specific." The problem the service unit facedhow to reduce
inventory while still guaranteeing that it could make field repairs within
the two to four hour windowwas unique to its business unit. Plus,
says Klimke, "you get things done faster when you do things smaller."
Network Appliance has 195 geographically dispersed service depots worldwide,
with 95 in North America. These depots are supplied by a main service hub
in Louisville, Ky., which in turn gets parts, such as disk drives and data
caches, directly from its factories. With a field parts planning module
from Baxter Planning Systems in place, Network Appliance spent June 2000
through July 2001 installing two pieces of quick-hit software from Fairport,
N.Y.-based Xelus and Fremont, Calif.-based WorldChain that would turn the
information in the Baxter system into inventory reductions.
After the Xelus demand-planning system comes up with a forecast, Network
Appliance uses replenishment-planning software from Baxter to determine
how much inventory it needs to keep at its depots and at its Louisville
hub. When a customer initiates a service request, it is routed through
WorldChain's supply chain execution software to the nearest depot, which
services the customer. The Baxter and Xelus systems then in turn automatically
tell the hub and the factory that they must ship out another part to replace
the one used. Now that the depots, and the hub, can rely on being expeditiously
restocked, they can carry minimal inventory.
Since the new system was implemented in July 2001, Network Appliances
has seen on-time service responses increase from 70 percent to 98 percent.
Furthermore, despite a 430 percent growth in its customer base since it
first started the demand-tracking project in 1999, its inventory level
has increased only 120 percent.
You Take It. No, You.
Until the mid-1990s, the airplane engine manufacturers (Boeing, GE, Rolls
Royce and others) that bought metal castingsrotating engine foils
and body parts for wingsbought generic parts and customized them
in their own shops. Then, in order to reduce their own inventory, Howmet's
customers began de-manding finished pro-ducts, sending How- met scrambling
to find subcontractors who had the tools and skills to do that. The Darien,
Conn.-based Alcoa subsidiary's cycle time doubled, and inventory turns
were cut in half.
In order to cut its cycle time back down from 10 weeks to five, Howmet
needed to improve communication with the small shops that do the final
casting customization. "If I make a casting and simply ship it to
a machining supplier, he may get up in the morning to find 100 castings
on his [loading] dock," says Prime Operations Director Kevin Mullowney.
"That creates delays." Mullowney realized that Howmet needed
to figure out how to let its suppliers know in advance what it would need
In December 2001, Howmet launched a supply chain visibility project using
software from Cupertino, Calif.-based Exemplary. The software extracts
timetable, quantity and parts-type information from Howmet's purchase
order system and automatically routes it to the appropriate supplier so
that it can plan in advance. The system cost $500,000 and went live in
Mullowney says that Howmet gets about 80 percent of the system's benefits,
principally in cycle time reduction, while its suppliers get the remaining
20 percent, as they can now plan for the arrival of Howmet's castings,
staff their shops accordingly and turn them around more expeditiously.
"Two months ago, I may have carried $10 of safety-and-buffer stock"
to guard against late deliveries from subcontractors, says Mullowney.
"But excess inventory is waste. By having better visibility, I may
be able to reduce that to $6."
Small Is Beautiful
The assumed advantage of enterprise software is that what it lacks in
specificity, it makes up for in breadth. But the CIOs in this article
learned that the trade-off isn't always worth it. "The majority of
people buy very large applications," says Juniper's Perdikou. "But
they are only getting about 15 to 40 percent of the functionality out
of it. You're better off buying something smaller and really using it."
Network Appliance's Klimke says that when you have a specific problem
you need a specific solution. If the problem is unique to a business unit
or involves a small number of business processes, then it probably needs
a targeted solution. However, Klimke suggests that CIOs make sure that
similar problems aren't popping up in other business units before going
small. "Requests that come from the business are typically narrow
in focus," he says. "They come from directors who have a narrow
responsibility. If we see requests with a common thread, we group them
together, and that is when you should end up with enterprise solutions."
Finding the right software is another challenge. Typically, business
managers have a long list of functionality they want the new system to
have. In order to keep the project small, says Perdikou, "you need
a truthful businessperson who will tell you where 80 percent of the benefit
is. All these other parts might get you 5 percent of the benefit and take
you a year to do it."
The more focused a small project is on a specific goal, the faster it
can be completed, the easier it is to get internal buy-in and the sooner
the payback. There is a danger to small projects, however, warns Klimke.
"You end up with a lot of solutions," he says. "And you
have to keep careful track of that." Otherwise you wind up with disparate
systems all over the placethe very problem enterprise software set
out to remedy in the first place.