How Verizon Flies by Wire
Two major shifts have changed the dynamics of information
hoarding and given CIOs the chance to finally make these systems work.
The old dream of
dashboards has new life. Here's how to make them work for you
Every 15 seconds, a new set
of charts telling the story of Verizon Communications' performance flash onto a
giant LCD screen in CIO Shaygan Kheradpir's office - everything from the
percentage of customer calls resolved by voice systems and how many repair
trucks are in the field to the resolution of IT systems problems. There are 44
screen-shots in total, and they cycle continuously, all day long, every day.
Kheradpir has seen them so often that they are burned into his memory. He can
tell at a glance when the lines on the charts aren't trending the way they're
supposed to. The system works much the same way for 300 to 400 managers at
every level of the company, says Kheradpir.
But when Kheradpir first
installed the system at the telecom giant in January 2002 (shortly after
becoming CIO), he didn't know that the fortunes of these dashboards had been
trending down for two decades at the company. After preparing a rudimentary set
of about 12 different business metrics for the private consumption of Lawrence
Babbio, vice chairman and telecom president of Verizon, Kheradpir marched into
Babbio's office, turned on an LCD monitor and asked him what he thought.
"He said: 'This is
great. Keep going'," recalls Kheradpir. "But then he surprised me by
saying that Verizon [and its predecessors] had been doing this for 20 years,
and it had always failed before. That was a cue to me to make sure that I course-corrected
on this thing as I went along."
Babbio was right to be
sceptical. Systems designed to give upper management a data-driven view of
what's really happening in their organizations have been around - and have
failed - for much longer than 20 years; they used to be called decision support
systems. They failed because the information was difficult and expensive to
gather, and therefore shallow and dated. And they failed because the systems
reinforced the traditional hierarchy of information in companies, in which top
managers hoarded all the important performance information and kept it from the
people who could take quick action on the numbers: line managers and employees.
Of course, line managers
and employees did their share of damage too. If they had information, it was
easy enough to conceal or delay that which reflected poorly on them.
"Information is power, and that's been the case ever since there was
information," says Laurie Orlov, a vice president of Forrester Research.
Two major shifts have
changed the dynamics of information hoarding and given CIOs the chance to
finally make these systems work. First, of course, is the Internet; it's no
longer prohibitively expensive to gather information that is timely and
distribute it widely. The second shift is the responsibility that executives in
public companies have, after the Enron disaster and the advent of the
Sarbanes-Oxley Act, to know about problems brewing down in the depths of their
organization. Dashboards can give top managers insight. "Ideally, the
dashboard would surface something before it became a material event so they can
fix it," says Orlov.
Thus, despite the futility
of Verizon's previous dashboards, the stage was set for Kheradpir's effort.
"No one argued against the case for dashboards," he says. But the
design was controversial. Different people at different levels had ideas on
what metrics mattered and who should see them. The system needed to be capable
of serving a vast audience, yet it could not be too complex or too simplistic.
Dos
and Don'ts of Dashboard Design
Like so many other
enterprise systems - ERP, for example - the technology is the (relatively) easy
part of a dashboard project. Verizon's is based entirely upon HTTP, and uses
software agents to gather information from the many systems around the company
(billing, for example, or phone line provisioning) and to bring it to a central
server where a master agent parses the data into tables. Agents on each
employee's PC are configured to ask the server for the data that matters to
them.
Then it was time to tackle
the organizational design. In a decision that probably saved the project,
Kheradpir did not try to gather consensus about metrics before rolling out the
dashboard. The first dozen or so metrics he handed to Babbio were based on
Kheradpir's knowledge of the business and his own hunches about what he thought
Babbio might want. He also decided not to give the metrics to anyone else at first.
"With things that are change-oriented, you just have to get people started
and then keep working it," says Kheradpir. Babbio immediately ordered up
other metrics and asked Kheradpir to tweak some of the ones he started with.
But that first version Kheradpir
gave to Babbio and other top executives had a fatal flaw: It required effort.
The executives had to log in and search for the metrics they wanted; usage
quickly fell off. Kheradpir began to fear that his dashboard dream would share
the fate of all the others before it.
Then he had a simple but
powerful epiphany. Initially, "people had to click and click to get where
they wanted", he recalls. "And pretty quickly, I saw that they were
too busy, and they just weren't going to do it. So we did a design that was
autonomous. It was in their face; they couldn't ignore it." Instead of
having to find metrics, the metrics would come to them - automatically,
constantly.
And that's when things took
off, in late 2002, says Kheradpir. The dashboard collection worked its way down
into the organization, as executives (who had it) called subordinates (who
didn't) to complain about metrics that the responsible parties couldn't see.
Kheradpir started getting calls from all over Verizon. Some were curious;
others were angry.
"People asked: 'Why
are you doing these measurements, and why are you publishing them to the
president?'," recalls Kheradpir. "And that gives you an opportunity
to open up a dialogue. Maybe they don't like the measurements or they think they
are being interpreted incorrectly. So I would say, OK, how would you do it? And
they'd say: 'Well, you'd better talk to X about that.' And I'd say, Great, he's on the team." Kheradpir recruited affected
businesspeople to help create or modify their own metrics for the system, which
is maintained by a small IT team.
But as usage spread
exponentially, so did the number of metrics. The dashboard server now contains
over 300 different metrics about Verizon's business, each of which can be
drilled into for more detail - much more than anyone could ever hope to read,
much less interpret. Kheradpir's solution is to have business unit chiefs pick
the metrics that matter most to them and their line managers, and dole them
out, giving managers access to only the metrics that their supervisors (and
Verizon's legal department) deem necessary. And the community that sees the
metrics is limited; 300 to 400 managers is a small percentage of Verizon's
200,000 total employees.
In a sense, it's the
traditional hierarchy of information reborn in an Internet skin. But Kheradpir
says lower-level managers get to learn much more about the business than they
used to, and they can appeal and request a different mix of metrics if they
feel short-changed or overwhelmed.
Manage
by the Dashboard Light
Giving line managers access
to metrics that matter is important, but most dashboards fail because the
systems stop there. They do not establish a clear link between cause and
effect, between numbers and action. Say that sales are down for the day. Why?
The metrics that could trace the cause and narrow accountability often aren't
present or aren't presented in a way that would let the user figure it out.
Kheradpir is trying to
change that. High-level metrics screens appear as a tabbed browser page. Within
each tab, there are drill-down links to narrower metrics. His metric screens
also carry the name of a person deemed responsible for that particular
activity, with contact information and a link to Verizon's instant messaging
system. It's a sure way to get people to use the system.
Kheradpir has a screen he
calls "The Wall of Shaygan" that shows live links to the more than
100 major IT systems around Verizon. Each system blinks green, yellow or red.
When Kheradpir sees a system moving into yellow, he pings the administrator
responsible. When he first set it up and the pinging started, employees
wondered what had hit them. "They all started saying: 'What is this Wall
of Shaygan and when can I get one? I'm tired of getting calls from Shaygan',"
he laughs.
Before calling, Kheradpir
often checks written logs that are attached to many of the IT systems, in which
IT employees can explain why metrics have gone out of line. Forrester's Orlov
says dashboard systems at many companies lack this kind of mechanism for
employees to respond to and explain problems.
Still, says Kheradpir,
there is a simple way to avoid getting calls: Keep your metrics in an
acceptable range. "If these metrics aren't moving the way you want them to
day in and day out, pinging the employees is the least of our problems,"
he says. "It's much better for us to jump in and try to fix it rather than
have an explosion much later."
Kheradpir says that all his
calls have raised morale in his department rather than lowering it. "My
organization is becoming more self-mobilizing around what they need to take
care of. Over the months I have pinged them less and
less. My organization is a lot more agile and we're all a lot more
together."
As with most knowledge
management systems, the return on the investment (which Kheradpir will not
reveal) is hard to pin down. So Kheradpir is measuring the acceptance level,
which is high among managers. "When I became CIO, I had a mission to make
this thing stick," he says. "And I'm happy to say that it has been
embraced more than I ever hoped for."
If Kheradpir's dashboard
effort had its own dashboard, it would be trending up - for the first time in
20 years.