Cannibalism Is a Virtue
In Computer Business,
Tandem's CEO Learns
To Beat Fast-Moving Rivals,
It Guts a Robust Product
By Unveiling Better One
Not a Moment Too Soon
By Joan E. Rigdon
The Wall Street Journal
(Copyright (c) 1994, Dow Jones & Co., Inc.)
James Treybig is a reluctant cannibal.
For two decades, the president and chief executive of Tandem
Computers Inc. built his company's fortunes on a simple formula: Design
powerful, refrigerator-sized computers that work only with one another
and sell them to corporate customers for the highest possible prices.
When revenue skidded last year, Mr. Treybig was backed into a corner.
His boss, Tandem Chairman Thomas Perkins, the biggest individual
stockholder, was considering firing him. His customers were screaming
for lower prices. In the end, Mr. Treybig was forced to adopt a strategy
that could save or sink the Cupertino, Calif., company he had founded 20
The strategy is cannibalism: gutting an existing product line, even when it
is selling well, by introducing better versions for competitive prices -- then
slashing prices on the old line.
The attitude is described succinctly by a successful practitioner at rival
Hewlett-Packard Co. "When you are at the top, you have to have the
courage to say, `I have to stop investing in this great product, and I'm
going to use the money to generate a new product that will kill it,'" says
Willem Roelandts, the head of H-P's computer business. "If you're not
doing that, let me tell you, there's going to be some competitor that's
going to do it for you."
This "we eat our young" tactic isn't entirely new. As far back as a century
ago, Procter & Gamble Co. began introducing new-and-improved
versions of its detergents to gain market share. But in the past decade,
increased competition and shorter product cycles have made
cannibalizing a means of survival, especially in the hypercompetitive world
of computers and consumer electronics. To keep up, high-tech
companies must constantly find ways to cut costs and prices while
marketing new products that eventually destroy their own previous
The goal: generate profits from volume and manufacturing efficiency, not
from artificially high prices that mask fat costs. "High prices are like
drugs," Mr. Treybig says now. "You can get hooked on them."
Giants such as Intel Corp., Motorola Inc. and H-P are accomplished
cannibals. H-P, which dominates the world market for computer printers,
introduced its black and white DeskJet 500 model in 1990 at $729. Even
as sales soared, H-P cut prices. In 1993, the company rolled out an
inexpensive new color model, eventually slashing the old DeskJet to
But most high-tech companies aren't good cannibals. Far more typical are
the likes of International Business Machines Corp. and Digital Equipment
Corp., which lost much of their business to smaller competitors because
they weren't aggressive in developing machines that would render their
The main reason companies delay and dither is a grim mathematical one:
Replacing a product line with one that is priced at a fraction of the
previous one means a company must cut costs and keep its work force
lean while ramping up sales. Often the new product is so different, or the
new price so low, that the company has to reinvent itself and find new
The question Tandem faces is whether a high-tech company that grew fat
on lack of competition can become a successful cannibal. As Mr. Treybig
has learned, making the switch at a rather late date is a harrowing and
wrenching process. Over the past two years, he has had to fire friends,
swallow pride and destroy parts of what he created.
And it won't get better. If Mr. Treybig succeeds -- he promises to come
out with low-priced, more-powerful computers every 15 months, and to
increase unit sales by at least 36% this year and next -- he will have
simply won a spot on a treadmill of unceasing innovation, anemic profit
margins and penurious cost controls. If Tandem can stay upright, it will
survive -- but it will thrive only if it can set the pace.
Mr. Treybig, 53 years old, started Tandem in 1974 with a $1 million
check from Mr. Perkins, a venture capitalist. By 1990, he was sitting
atop a business with $1.9 billion in annual revenue and a solid list of
He got there by selling a hot product that his customers told him they
couldn't live without: "fault-tolerant" computers, designed to keep
working and protect data even during power outages or natural disasters.
Banks, telephone companies and stock exchanges stake their businesses
on Tandem computers' ability to process millions of transactions, 24
hours a day, without a glitch. A single Tandem failure, even for an instant,
could paralyze thousands of automated-teller machines or derail an entire
city's phone calls.
Mr. Treybig enjoys pointing this out. "You better pray for us. The day
we're not here, the stock markets of the world stop," he recently told a
group of hospital executives in Seattle.
Playing on that fear, and on the fact that his multimillion-dollar computers
worked only with one another, Mr. Treybig charged hefty premiums for
his products, and got away with it. Throughout the 1980s, Tandem
enjoyed profit margins as high as 19.4%, on par with then-powerhouse
Tandem's sales force had one of the richest compensation plans in the
industry. Salesman Pete Engel recalls buying a new Mazda RX7 every
year. "We used to be able to sell one machine a year and have a pretty
nice living on it," he says.
It is at this point, when profits are plentiful, that successful cannibals
introduce the next generation for a slight premium and slash prices on the
older line. But Tandem didn't. Woozy from success, it expanded into new
markets, where fault tolerance wasn't needed and wouldn't command a
premium. Instead of cannibalizing, it kept prices high, figuring customers
would be mollified by excellent quality.
While competitors were beginning to build their systems to run Unix, a
popular operating system, Tandem stubbornly stuck to its proprietary
system, called Guardian. The advantage of Unix is that it is "open" -- it
runs on many brands of computers. A customer who writes a program
for Unix on one computer can run it on any other computer running Unix.
The customer can switch to cheaper computers as they hit the market
without losing its investment in software.
But a program written for Guardian could only run on Tandem
Tandem's first warning came in 1989, from a major long-distance
company that had bought $100 million of Tandem gear over the years.
Two of the customer's senior vice presidents told Tandem their company
had decided to write its future software for Unix. If Tandem couldn't run
Unix, the company would take its business to DEC.
"We were pretty shocked," says William Heil Jr., a Tandem executive.
Meeting with the phone-company executives, the Tandem team tried to
talk them out of using Unix. But the guests began screaming, asking how
Tandem could be so blind, Mr. Heil recalls. Before the meeting ended, he
says, "we were screaming at each other."
Rattled, Mr. Heil and his colleague discussed whether it was possible to
meet the long-distance company's demand. One scenario was to make
Guardian open, so it could run on any Unix machine.
Mr. Treybig's response: "Open is death." Running Unix would make
Tandem computers a commodity, he complained, forcing the company to
compete with many others on price and features.
Tandem chose instead to go halfway: It made a low-end computer that
could run Unix. High-end machines would continue to run Guardian.
But Tandem's biggest customers, who use high-end machines, continued
to clamor for Unix -- and began replacing Tandem computers with
models from H-P, DEC and others. In 1991, one of Tandem's biggest
accounts, Wells Fargo & Co., threatened to defect.
That year, Tandem reported its first-ever quarterly loss. But rather than
redesign its computers and cut prices, it took another half-measure:
Chairman Perkins walked Tandem's hallways, soliciting executives for
names of people to fire, and gave the list to Mr. Treybig, who
successfully defended many on the list. But he reluctantly agreed to cut
the work force by 6%, about 700 workers.
Mr. Treybig says he cried over some firings. But he also felt bad about
not having done some earlier. "What I realized is that if I didn't do my
job, no one would have a job," he says.
Tandem's financials brightened briefly. But by early 1993, it was clear
that cost-cutting alone wasn't enough. In the third quarter, year-over-year
revenue growth -- once in the double-digits -- shriveled to 2%.
At the time, Tandem was developing a new computer, code-named
Himalaya, with twice the power of the Cyclone, then the high-end model.
A good cannibal would have immediately announced the new product
and slashed prices on the old one. Instead, Tandem decided to continue
to milk the Cyclone until the Himalaya was ready to ship in late 1993.
When Mr. Treybig revealed this timetable to META Group Inc., a
Westport, Conn., research firm, in the spring of 1993, he got a tongue
lashing. The delay in slashing Cyclone prices "sounded like a voice from
IBM's playbook four years ago," says Nili Young, then a META analyst.
"He wasn't in a position to wait. His customers weren't in a position to
wait." She warned Mr. Treybig that two big London banks planned to
stop buying from Tandem as soon as they could.
The meeting finally pushed Mr. Treybig to gut his old product line and
embrace cannibalism. Breaking with company tradition, he announced the
Himalaya immediately, five months before it existed, promising that a new
design would allow prices of one-third to one-sixth those of comparable
machines in the older line. And by late 1995, the machines would run
either Unix or Guardian.
To keep the older Cyclone line moving, Mr. Treybig slashed prices and
offered special discounts to Cyclone buyers who wanted to upgrade to
To make up for the lost revenue, Mr. Treybig cut every employee's pay,
including his own, by 5%, and nixed merit raises. Steeled by the 1992
layoffs, he savaged the family culture he had spent two decades nurturing
and made plans to cut 1,500 more jobs by 1995. The cost: a
restructuring charge of $451 million.
He had little choice. "You can't go on just slowly dying, you have to fix,"
Mr. Treybig says.
Convincing his sales force of the need to cannibalize was tough. He was
asking them to sell twice the number of computers for lower commissions
and pay. Several sales representatives tried to convince Gerald Peterson,
general manager of sales, that there was no pressure to cut prices. At a
sales meeting in Europe, Mr. Peterson recalls, representatives told him
the cuts were "a crazy American idea."
Mr. Peterson told them, "If we don't do this, we are going to be out of
business." As for compensation, Tandem salespeople remain among the
highest paid in the industry, earning $100,000 to $400,000 a year, he
Tandem buttressed its product strategy with new ads. Out went the
splashy but cryptic customer testimonials, replaced by plain ads
comparing costs between Tandem and competitors.
The strategy appears to be working. Customers now can buy a low-end
Himalaya for $339,000 -- when a low-end Cyclone sold for $900,000
six months ago. And a high-end Himalaya sells for $700,000, compared
with $4 million for a machine of the same power a year ago.
Because of the price cuts, Tandem now is being considered for uses it
wasn't even asked to bid on before. Rabobank NV in the Netherlands,
which was about to cancel an order for Cyclones when Tandem slashed
prices last year, is now considering buying a Himalaya. In the fiscal third
quarter ended June 30, Tandem added 76 new accounts, including
Telecom Securicor Cellular Radio Ltd., a British wireless concern, and
Pick 'n Pay Stores Ltd., a major South African retailer.
In the third quarter, Tandem sold more than 500 of its high-end
Himalayas, about double the number sold in each of the first two periods;
it hopes to have profit margins of 8%.
The result: Tandem is promising record earnings this fiscal year (the
current record is $1.17 a share in 1989). Internally, the bar is higher. Mr.
Treybig calls this fiscal year the "$1.30 year."
Looking back on the angst it took to get to this point, Mr. Treybig
concedes he should have moved faster. "There's no question we made a
mistake. We would have been better off to go for low-priced systems."
In Europe, where markups were especially high, he says, "we were
raping the customer."
If Mr. Treybig can pull off this cannibalization, Tandem would be the first
old-line computer company to reinvent itself successfully. But if he fails,
he could lose his job, and with it, the only life he has known for 20 years.
Says Chairman Perkins: "We're not married to Jimmy."
The battle is hardly over. Customers are pressing for even lower prices.
And while Mr. Treybig dallied, competitors made big inroads. Wells
Fargo, for instance, bought a lot of Hewlett-Packards. If Tandem
"delivers what they say they're going to deliver, they'll be competitive,"
says Barry Lynn, Wells Fargo's executive vice president of customer
information. "Will they be able to break the existing strongholds of some
of their competitors?" he continues. "That would take a fortuneteller to
Back to Joan's clips