One of my earliest lessons learned was to
avoid creating the appearance of competition between traditional financial
measures and their new non-financial siblings. It therefore seemed logical to
integrate the scorecard into Analog's existing financial reporting system.
This took on two forms: inclusion in the "Red Book" and Analog's
online Executive Information System (EIS) called "IMAGE." The
Red Book was the quarterly financial report (in a red binder, of course) that
was prepared principally for the Board of Directors. IMAGE was a mainframe
based data warehouse and graphics system, used almost entirely by headquarters personnel, that could be accessed through hardwired remote color
terminals.
Starting in February, 1988, we included in the
Red Book and IMAGE our corporate and divisional scorecards as well as our delivery performance
metrics. In a January 20, 1988 memo announcing this change, Analog's Corporate
Controller (who was also a member of the Corporate QIP Council) explained:
"The Red Book will also be the
medium for hard copy distribution of the QIP measures as they evolve.
Initially this will mean displaying the improvement curve [n.b.
half-life curves] for the on-time delivery measure for customer service which
is our only implemented measure so far.
The final section will contain the entity
scorecards or quarterly performance audits. These track the key measures
of success on a quarterly basis with comparison to the benchmark goals for
each measure. This report is designed to fit on one page and to provide
for comments as to the reasons for missing the goal."
But the IMAGE system had five major
drawbacks:
-
Since it was a mainframe based, dumb
terminal system, people who were not hard wired to the mainframe could not
access it in graphics mode.
-
It was very slow and user
unfriendly. Navigation required mastery of a large number of indecipherable
acronyms and command line instructions.
-
Hard copy printout required special
hardware.
-
It was written in APL, which
required the use of an outside consultant for all program changes.
-
It was not easily movable to a lower
cost hardware platform (a minicomputer, at that time).
I therefore successfully (well maybe not
so successfully, as you'll see) spearheaded an effort to purchase a
commercial software package that would mitigate IMAGE's drawbacks. A
careful search led to the selection of Pilot Executive Software's Executive
Information System (EIS). PILOT maintained all data on the mainframe but
the bulk of the graphics were done on local PC's which could connect to the
mainframe using modems over local phone lines. By the end of 1988, we had
completed the transition from the IMAGE to PILOT delivery platform. To
the user, the transition was transparent with regard to content and the various
displays. However, navigation was now by mouse click and hardcopy was
generated at the PC's printer.
I've recreated a demo of the PILOT system
(see: PILOT demo
and recommended viewing
note) which allows you to view examples of all of the information that it made available. It is identical to
the original except that all possible hot buttons are not active.
Clicking those outlined in yellow allows you to realistically navigate as in the
original. Several rules were maintained throughout:
-
Variances were
noted throughout the entire system using a color coding standard: a result
shown in red represented an unfavorable variance, while a green entry denoted a
favorable variance.
-
Clicking on a number produced a
graph of its time history.
-
Clicking on the name of a metric
produced a table of its disaggregation.
-
Clicking on the name of an entity
produced its relevant scorecard.
-
All metrics were defined so that a
decreasing y-axis value represented improvement.
-
3s
control limits were placed on all appropriate charts to define the region of
statistically insignificant month-to-month changes. The most recent
three data points were colored red or green depending on whether they were
above or below the control limits.
-
The improvement half-life was
calculated and appeared at the bottom of all time histories.
Clicking on the half-life produced its time history, our principal metric
of process learning.
-
Clicking on a statistically
significant variation (red or green colored) produced a commentary,
written by that metric's owner, that identified root cause,
corrective action, accountability and timeline information. These
commentaries remained active until the variance was eliminated.
Plans for expansion of the IMAGE/PILOT
system included drilldown to lower level scorecards (the system only drilled
down one level below the corporate scorecard) which would be inputted and
maintained by its owner. We also planned to link PILOT to Lotus Notes for
further expansion of the scorecard commentaries and integration into our meeting
agendas and agenda item preparation materials.
Unfortunately, in March of 1990, while I
was away on a business trip, the Executive Committee decided to terminate any
further development of PILOT under the guise of a cost reduction effort.
My view of what happened does not belong here, but it's in the queue for a
future Art of PM essay which will be entitled "The Politics of
Information."
Ironically, PILOT's death was concurrent
with its birth outside of Analog. In a February 1990 interview that
appeared in ComputerWorld, I described our adaptation of a traditional financial
EIS to non-financial metrics and the balanced scorecard and my expectation that
it would eventual replace our traditional financial-only management
system. Shortly after that, I was invited to be a plenary speaker at the
Institute of Management Sciences DSS-90 conference (see: May
1990 SS-90 Presentation) held in May, 1990.
One of the other speakers at DSS-90, Jack
Rockart, the Director of MIT's renowned Center for Information Systems Research
(CISR), came to visit me on June 21, 1990 for a first-hand demo of PILOT.
Jack was often referred to as the "father of EIS" so his visit and
compliments were much appreciated and helped mitigate our posthumous blues.
Jack invited me to speak at a CISR seminar held in December 1990. A month earlier, I made a
similar presentation at a seminar sponsored by Decision Support Technology, Inc..
PILOT's final tribute came in the fall of
1991 with Bob Kaplan's acceptance of my offer to set up a full working version
of our system on the Harvard Business School's computer so that students could
use it in conjunction with the Analog Devices case that he had recently
written. David Friend, CEO of Pilot Executive Software, agreed to provide
support and a free copy of his software. The three-way team of ADI-HBS-PES
personnel did a masterful job of creating a working demo, which from all reports
added significant value to the student's learning experience.
I have recently seen a number of demonstrations and
actual applications of current balanced scorecard support software and I would
welcome comments about how they have improved over our 1990 PILOT
implementation.