Skunkworks as Distributed Networks
by Jessica Lipnack and Jeffrey Stamps
The Seybold Series #3 - August, 1989
Perhaps the best known model for innovation, "skunkworks"
came to prominence in the 1980s through two of the decade's best
sellers: Peters' and Waterman's chart-topper, which also made "excellence"
a keyword in the business vocabulary, and Tracy Kidder's The
Soul of a New Machine.
In Kidder's book, technology is the main character, and it is conceived
and born in a skunkworks. According to In Search of Excellence,
"a classic skunkworks consists at any one time of eight to
ten people...located in a dingy second-floor loft six miles from
Ten years ago this was an accurate description of how development
work was done. Today, however, a different model for innovation
is possible because of computer networks. Instead of one "second-floor
loft six miles from headquarters," now companies have many
such operations, linked by communications. Skunknets are
the new distributed work dynamos.
Clipper: World-class Product, Record Revenues
One such example is a highly successful technology development
project, carried out in the new mode: globally distributed groups
linked by numerous media working on the same project, code named
A worldwide project, Clipper produced a world-class product built
at a low cost in record time with record revenues.
In just 15 months, a core group involved 45 people in more than
a dozen locations with several thousand more participating indirectly.
Within ten weeks of the product announcement, 1000 machines were
builtóand shippedófrom three locations separated
by an ocean.
"People who worked on this project say they will never forget
it," says the project manager.
How did they do it? For those who believe high performance can
only come from people packed together in one place, Clipper is an
excellent example of how exceptional work is done in a distributed
environment, in a network.
Clipper successfully integrated both organizational and technology
networks. Each node in the network had multiple links to
the others, held together by clear shared purpose.
This is a network that worked.
The Nodes Know
The project was distributed in two important ways. First, it was
developed with the participation of design, engineering, manufacturing,
shipping, marketing and sales functions. Second, it involved multiple
Seven company locationófour in the U.S., two in the Caribbean,
and one in Europeóprovided the inner circle of the project.
Four secondary manufacturing plants included three in the U.S. and
one in Canada. Yet another U.S. location kicked in computing resources
at-a-distance at a critical point in the project. Finally, two external
organizationsóa key vendor on the west coast and a key licensee
on the eastóbrought the number of nodes in the network to
With people scattered across tens of thousands of miles, how did
they possibly achieve success? The answer is simple: They communicated.
Extensively. Using many media.
First of all, they had a regular schedule of face-to-face meetings,
an important lesson for those who believe that online communication
replaces "the real thing." The project manager expressed
this well when she said, "You don't build trust over the wire.
That takes flesh-and-blood" but it doesn't have to take a long
time to make it happen."
The entire project team of 45 met monthly ("The only excuse
for not coming was if your mother died") and the four team
leaders who formed the project management group met weekly ("But
we checked this weekly; if we didn't need to meet, we didn't").
In between, people traveled ("There are a lot of project miles
on these people").
Everyone used electronic mail ("Key people have to commit
to read mail in real-time). Computer conferencing was used to work
out technical problems. ("Engineers don't talk. They type.")
But of course they do talk, and telephone was used extensively,
even more so after a separate satellite dish was installed in the
Caribbean so that locations there could bypass their local already
stressed communications system.
Nodes and links alone do not a network make. A network needs purpose,
the glue that bind, and Clipper's was clear: build a break-through
computer by leveraging existing technology to fit a clear market
Easy enough to say. But for a distributed group to work effectively,
especially on a complex project, everyone needs to share the same
model of its work.
Clipper accomplished this by creating a common design database
that helped bridge the traditional "throw it over the wall"
mentality that separates engineering and manufacturing. Eliminating
the stale paper problem, the database enabled manufacturing to cut
its engineering response time from weeks and months to a matter
of hours. Distribution lists were built into the mail system such
that the "right" people received the "right"
information "right away." ("With paper, you could
never be sure that what you were working on was current.")
The database also enabled the key licensee to do a simultaneous
release of a different version of the productóthe first of
its kind as well. ("We had fewer meetings than we expected.
Mostly what we did was ship them data.")
For a project with multiple centers, multiple leadership is the
only answer. In this regard, Clipper excelled. Each of the four
project managers had two reporting relationships: an indirect one
to the overall project manager, and a direct one to their local
site managers. Each of the remaining sites followed a similar pattern,
with the local manager reporting into the site as well as reporting
of one of the four project managers.
In some locations, new reporting relationships were established,
acknowledging the importance of the project. One manager in a manufacturing
plant, for example, was elevated to a position on the plant manager's
staff, a role that is usually buried several levels down in the
organization. Such reorganization proved critical in meeting deadlines,
since signature loops and approval times were thus substantially
shortened. ("Everything came down to the schedule. When we
needed to make up time, we did more things in parallel.")
Benefits of the Clipper Model
Clipper's report card tells the story. Clipper met its cost, time
and revenue goals. Indeed, it exceeded them. Just a fraction of
the first quarter's earnings paid back the development costs, as
revenues continued to rise.
But those were not the only benefits. The distributed project team
leadership was able to innovate in a way that could not have been
done if everything had been under central control.
When a bottleneck developed in the shipment to the Caribbean of
the sheet-metal boxes that housed the computer, the local project
leader, who knew the other business people on the island, cut a
deal with the local overnight delivery service. "Headquarters,"
thousands of miles away, would never have known about this option.
In the same way, immeasurable time was saved by being able to start
up the project without having to move a single person or disrupt
a single family. The project was off the mark instantly.
"A project like this gets momentum and there's no stopping
it," the project manager concludes. "There's a point where
the product becomes perceived as successful [and] it gets really
exciting...This product had lots of synergy."