An interview with Ellen..This is one sharp lady,IMO
The Start Up and the Grown Up -- A 30-Year IBM Veteran, Ellen Hancock Gets
Young Exodus Down To Business.
Dec. 03, 1999 (LTH - CMP via COMTEX) -- Every startup has to grow up
someday-even an Internet startup. Often, the passage begins right around
the time of a company's initial public offering (IPO). A seasoned executive
is hired to lend management skills to the company's visionary founders-sometimes
eventually to replace them as they move on to their next vision. The cynic
could call this window-dressing; adding a star to the executive roster
can help make an IPO soar. But realists see the IPO as more than a one-time
payout, which leads them to scoff at the notion or worth of trophy executives.
For them, an IPO is just one step in an ongoing campaign to convince investors,
customers, partners and other executive talent-day in and day out-that
your company is the one for them. It represents a time to button down
and turn an Internet vision-thing into something that looks a lot like
routine operations.
Ellen Hancock may fill both the cynics' and realists' visions. She joined
Exodus Communications Inc. (Santa Clara, Calif.), which runs networked
Internet data centers and services, nine days before the company went
public in March 1998. What's more, she joined as company president in
the middle of its IPO road show. Star power? She had it; she's instantly
recognizable to many from the conference circuit and pages of the business
press. Buttoned down? How could she not be after nearly 30 years at the
notoriously conservative IBM, where she managed 15,000 people? Most of
her time at IBM was spent on networking, with responsibilities for managing
systems like Ardis-an early mobile data network that IBM ran with Motorola
Inc.-and the global data network that IBM later sold to AT&T. Too buttoned
down? Maybe not. After IBM, the 56-year-old Hancock also had a stint at
the more freewheeling Apple Computer Inc. (Cupertino, Calif.).
Hancock, who was named CEO six months after joining Exodus, says she has
brought business processes from both IBM and Apple to bear on the startup,
many of them fairly basic.
In doing so, she has given Exodus the chance to continue its
evolution from its origins in 1994 as an Internet service provider (ISP)
to arguably the world's leading Internet hosting company, to an organization
that sells outsourcing, colocation, managed services and professional
services focused on hosting "mission-critical" Web sites-like Merrill
Lynch's new investment site. Along the way, she has bought a handful of
companies to add such expertise as Internet performance measurement (Service
Metrics Inc.), for instance, or to expand into Asia (Global OnLine Japan
Co. Ltd.). She has increased the number of data centers to 22, as of the
end of this year, and nearly quadrupled the staff, to 1,300.
Hancock has also changed the makeup of Exodus's customer base,
shifting it from mostly "dotcom" companies to include more "brick & mortar"
companies that are moving operations online. Through it all, Exodus has
grown revenue in excess of 40 percent for 11 consecutive quarters. And
its stock price at the end of November was pushing $130, up from around
$8 a share in January.
Only about half of the original Exodus employees have stuck with Hancock,
she says. "Some of the key people did say, once the company went public
and started to grow, that some of what they liked in a startup was missing,"
she says.
They may be right. Even Hancock admits it isn't the same old Exodus: "The
company has changed. We are bringing in people who have worked in more
of a corporate environment rather than a startup because we are no longer
a startup." And yet, she says, "Exodus still has a way to go in terms
of maturing as a company." It still loses money, for instance, with a
net loss of $24.7 million for the third quarter on revenue of $67.6 million.
Likewise, the Internet data center market that Exodus helped invent still
has some growing up to do. The young market caught fire this year as powerhouses
like AT&T and Intel Corp. announced intentions to build data centers worldwide.
The grown-up talked to tele.com recently about her startup.
---
Things are much more dynamic in this world than they were at either Apple
or IBM.
The people in the telco environment do not understand data centers. They
build the buildings wrong.
In terms of development, is Exodus a child? A teen? What is it? We're
a teenager. We are definitely a teen. We have not been to college yet.
Can you integrate a process-driven kind of management style with an Internet
hyperdrive style and maintain the best of both? Yes. In fact, you almost
have to. You have to be careful that whatever process you adopt does not
prevent you from making quick decisions. On the other hand, as you are
putting a product through development, it's nice to know that you actually
have some safeguards and have tested it and know exactly where you are
in the process. That helps your planning.
Why? What's at stake? I think several of us who come to Exodus from other
companies bring in an understanding of what can happen. And therefore,
we get the company prepared to not let it happen. When you talk to young
entrepreneurs and ask them to make a decision, they'll make a decision,
but it's not based on any background. It's just based on what they think.
One of the advantages an experienced executive brings is that we are basing
it on having seen some of this before. We have built up an instinct because
we have seen something close to this 20 years ago, 10 years ago, five
years ago-maybe in a different setting, different environment, but we've
seen it before.
Can you give me an example? In fact, some CIOs in enterprises are actually
nervous about their own Web groups because they're all 23 years old and
have never seen these problems before. And so the CIOs come to me and
say, "Do you have people who are older than 23?" And I say, "Yes, I have
quite a few of them over here." It's almost like they're looking for someone
who at least knows what could happen. For instance, we decided that we
wanted to put a tape backup system in all of the data centers because
I was personally nervous about how we were handling it. Before I came
here, we would send one of our engineers out to the customer's cage, and
then we'd be physically handling these tapes, and I said, "That's dangerous."
How has the culture of this company changed since you came in? When I
came to the company, it was small. The notion was that everybody, particularly
at the top, could do everything. If there was a problem, they would all
work on it. We had an off-site meeting to say, "Look, the heroes now are
all dead. There are no more heroes. There are, in fact, people who have
functional responsibilities and they are the people who are going to do
the work." So we had to learn about
delegation, which was not something the company had focused on earlier.
The other thing is that, before it went public, the company had a very
good reputation for problem solving. When a problem occurred, it reacted
very fast. After we went public, some of the customers said, "I understand
that you react very fast. How would you like to prevent the problem the
next time?" And so we went into prevention mode. When I started asking
about root cause analysis, that was not something the company had really
focused on.
Can you give another example? We have maintenance windows about twice
a week. So I asked how we decide what happens in the maintenance window, and the
answer was, Whatever is ready. And I said, "Is that really the right way
to do it? It may be ready, but perhaps it's a little bit dangerous to
have too many tasks in the same maintenance window. And do we tell the
customers what we are doing in the
maintenance window?" No. "So they can't tell if we are just cleaning the
rugs or changing a UPS system, is that right?" Uh, yeah. So we put in
a whole change-management process and we have a weekly meeting where the
VPs get together to agree what's going to happen in that
maintenance window. We tell the customers what's happening, by data center,
so they know exactly whether they should be planning for something or
relaxing over the weekend.
How do you approach these problems? In some ways we are trying to pick
up the best processes from both IBM and from Apple. For example, the way
we go through all of our development is now based, I would say, pretty
much on an IBM process. The way we are talking about our plans-our fall
plan and our spring plan-is very much similar to an Apple process. We
try to select which ones match most closely to what we want. Because Apple,
even though it is a large company, still acted like a startup in many
ways. They made a lot of decisions with no data. And on the Internet you
have to make a lot of decisions with no data, because you don't have the
time to get it. A lot of the Apple processes work fairly well for Exodus.
How hard is it to implement this kind of change? There's a lot of work
to be done. And when people say to me that competitors can come in and
do what we're doing, I say, "Yes, but they have to do what we just finished
a year and a half doing. They have to build up those same processes. They
have to build it across their whole culture."
Having come from 30 years at IBM, and then Apple, and then an
Internet company, how has that changed you? First of all, things are much
more dynamic in this world than they were at either Apple or IBM. You
have to get on the system almost every day and find out who announced
what for the day. Someone's buying somebody. Someone's announcing an IPO,
an alliance. You are much more sensitive to daily events. Maybe it's because
I'm a CEO, but you are much more sensitive to the stock market. Then,
because there is so much interest in the Internet right now, it's sort
of fun being in the middle of that buzz.
Speaking of being in the middle of a buzz, a lot of companies are moving
into the data center market right now. We picked to serve the market of,
essentially, the mission-critical sites. Very few people have come into
our part of the market. There are quite a few more people coming into
the intermediate space.
Could you define your market? First of all, it's customers who have a
fair number of servers. It would be like Hotmail, with 1,500 servers.
It's customers such as Merrill Lynch, which is bringing their brokerage
business online. The amount of money that kind of customer pays to their
hosting partner is quite a bit more significant than they might have paid
if they were in the intermediate market. The other
distinction between the two markets is that in the intermediate market,
the vendor often just fills the building with equipment. Then, when the
customers say they want two machines, they are given the serial numbers,
and they hardly ever go to visit the data centers. Whereas with ours,
you'll often find the customers wandering around the data center and,
essentially, working in their cages.
What do you give your customers? We are very comfortable with a customized
model, which means we will have a procedure-and agree to a procedure-with
Hotmail. We will have a different one with Inktomi. We will have a different one with Merrill Lynch. We will
have a different one again with Lycos. We tie in consulting and managed
services. The intermediate players are using a cookie-cutter, shared model
versus a customized model. And there's plenty of business for the two
models.
At a recent Bear Stearns conference, there were a lot of competitive local
service providers saying they had to get into the data center business.
Yes. They all want to be in on it.
How do you see the market developing? I think you'll see people coming
in, but I think some people are going to run into trouble because they
don't have the skills. The people in the telco environment do not understand
this environment. In fact, they build the buildings wrong. They build
them for telco equipment. The building, power, distribution: Everything
else is different when you are doing server farms and not a central office
switch. I expect to see more of them. But I don't expect to see a whole
lot of them being very successful. I think they will either be bought
out or give up.
From the demand side, where is this market in terms of its growth? In
the third quarter, we added four and a half customers a day to our system.
And the average price that the customer is paying went up. It's $176,000,
annualized. And our sales people are saying, "Look, we're leaving some
deals behind because we just don't have enough people to work on the RFPs
[requests for proposal]." So we just keep adding revenue. And so the analysts
have now increased the estimates for 2000 and 2001. For the first time
now, the analysts are showing us at over a billion dollars of revenue
in 2001, with about $600 million for the next year. So there are several
aspects that just say this business was even better than we thought it
was going to be about a quarter ago. And it's clear that we're the leader.
We've won somewhere between 75 and 85 percent of our bids since we went
public. We host 36 percent of the top 100 Web sites.
If you look out three years, let's say, how many Internet data
centers do you think there will be in the world, and how many will be
yours? Some people are getting concerned that there's going to be a glut
of data centers. If we have a concern, it's that we don't have enough
data centers. In London we opened 50,000 square feet. Before we even opened
the doors, we were building 100,000 square feet next door. If anything,
maybe we'll buy some data centers later if there is a shakeout.
---
ASP Watch
Ellen Hancock keeps a close eye on the nascent market for application
service providers (ASPs). If they proliferate, these providers-which operate
"rent-an-app" businesses over the public network-will be populating Internet
data centers like Exodus's. So far, though, they are a very small part
of the company's business-and a much smaller phenomenon than many expected
only a short year ago. Here's Hancock's take on ASPs.
Why isn't the ASP model growing as quickly as expected?
We decided around April that we would address the ASP market because it
seemed like an intuitive market. There was also a lot of discussion that
the market would really have taken off before the year 2000 because customers
would want to rent an application rather than install an application.
Because of the Y2K bug?
Yes. Now we're just watching everyone's experience-our own as well as
other companies'. We continue to see announcements by companies like Oracle,
that they're going to go live. But we have not seen the adoption rate
by customers as quickly as some people in the industry expected.
Why?
We've all been caught with intuition before. When I was managing the Ardis
network with Motorola, I thought wireless data netw
|
