The Half-Life of a Data Quality Center of Excellence
|
| Frank Dravis |
|
If you want to create a sustainable culture of data quality
in an organization and you want to grow the culture beyond the
bounds of one or two functional units then you will need an
enterprise data quality center of excellence (COE). The important
points here are sustainable culture of data quality and enterprise
wide. If your goal involves something less than those two requirements,
then a COE is not needed. For example, if you are starting your
first project and planning to cleanse two or three columns (maybe
some address or name data), and even considering a second project
later on, a lone project manager with the help of one or two
temporary analysts will suffice for your management needs. However,
the issue quickly becomes one of size and scale. A data quality
project manager or enlightened sponsor is all that is needed
to foster and sustain data quality inside a single function
or smaller business unit. But if you want to grow data quality
practices beyond the silo wall, you'll need a COE.
I've worked with five different clients who each had a
COE, including:
1. A defense department agency with a COE of 20 plus individuals
2. A high tech manufacturer of test equipment with a COE of
five plus individuals
3. A manufacturer of computer power suppliers with a COE of
six plus individuals
4. An industrial equipment subsidiary of a global conglomerate
with a COE of 40 plus individuals
5. A telecommunications service provider with a COE of nine
plus individuals
While each organization comes from a different industry and
has different revenue models, their COEs have a number of common
characteristics:
• Is a formal, permanent organization with a charter.
• Is managed by a data quality champion, usually a director
or manager of data quality. This person fervently believes in
the value of quality data.
• The director has strong communications skills, can build
and present a strong business case, and has latent sales abilities
(though they may not admit to it).
• Is staffed with a combination of skills, with the people
coming from both the business and IT sides. All personnel have
formal data quality job titles.
• Practice a clear application of data stewardship, which
is they structured their staff in line with sound data management
principals, and the roles and responsibilities of data stewards.
• Have a business line sponsor, usually a senior manager,
who the COE director works closely with.
• Most have immediate access to an outside consultant
they call upon for aid in employing best practices, but more
importantly, as an independent party to validate their findings
in front of senior management.
• And finally, all but one - which would become
a crucial weakness - has a steering committee or a data
governance board.
The data governance board is typically comprised of senior managers
from various line or business units in the enterprise. The board's
primary purpose is to be the final authority on all things data.
Many decisions concerning corporate information never reach
the board, just like many court cases never go to the Supreme
Court. However, information issues of crucial importance are
submitted by one or more board members at the behest of their
direct reports. Beyond the board's primary purpose, the
board serves a number of roles, and one role in particular is
to serve as a bridge across organizational change. In reviewing
how the five COEs evolved, the emergence of the data governance
board served as the tipping point. Prior to the existence of
the board, data quality was pursued on a project-by-project
basis within one functional silo. With the advent of the board
and its cross-enterprise membership, data quality processes
and practices propagated from the root entity into the other
business units whose managers sat on the board. The creation
of the board served as the catalyst to grow the series fledgling
projects into an enterprise-wide initiative – the COE.
In the case of the one organization that did not have a data
governance board, its sponsor – who was chartered with
a corporate initiative to optimize a key process (the order
to fulfillment process) – had the authority and political
capital to grow the COE beyond project-specific bounds. However,
as mentioned earlier, the board serves multiple roles, and a
key role is being a bridge across organizational change. When
and if the COE loses its data quality champion, the board is
there to appoint a new one. If the program sponsor is transferred
or leaves the firm, the board is there to select another from
its ranks. If a new business operation is started or the corporation
changes in some fundamental way, the board is there to issue
instructions to the COE program sponsor and champion, advising
them in how to adapt. With six, seven, or eight members on the
board, it could lose one or more members and still function
until replacements were found, thus sustaining critical mass.
This was sadly not the case for the firm with a COE but no data
governance board. Within four months the senior manager who
served as the COE program sponsor was transferred and the project
manager (the data quality champion) left the company. Much rhetoric
was given to finding a replacement for the champion, but with
the sponsor gone and no other senior manager lining up to support
the initiative it died. Within a month all but the most baked-in
data quality practices and projects were dropped. Regardless
of all the benefit and payback generated by the various data
quality projects up to that point, the culture at that firm
was not entrenched enough to overcome the loss of key personnel.
I remain convinced that a strong data quality initiative and
the building of a center of excellence can take place without
a board of governance. I also remain convinced the half-life
(decay rate after an implosion) of either is about the same
time for a humming bird to migrate south for the winter. And
from my place in Wisconsin that's about a month. The message
being, if you've put your heart and soul into building
a center of excellence, once you are done, take the time and
care to build a data governance board to ensure the COE's half-life
will never be measured.
Frank Dravis is Vice President of Information Quality at Firstlogic
Back
to Articles
Contribute
your own articles |
|
"...an invaluable
compendium of information on the most pressing challenge in
business today"
Sean Kelly
Comhra |
 |
"At last! Something
on data even an old Luddite like me can understand!"
Drayton Bird |
|
"...a really useful resource to understanding all issues relating
to data management"
Prof Derek Holder
Institute of
Direct
Marketing |
|
"...a down-to-earth, no-nonsense site designed to cover the
new practices and theory of Information Management"
Simon Lawrence
Information Arts |
|
|
