BI Project Success
- By Graham Tracey, James Riha
There is no one-size-fits-all strategy for business intelligence
project management. Instead, smart BI project managers will exploit
any number of approaches during a single project lifecycle.
OCCC's Institutional Intelligence team members faced hurdles throughout their BI initiative, but
they adjusted and learned to focus on outcomes and results instead of rules and hierarchical roles.
MANAGING BUSINESS INTELLIGENCE (BI) projects
in higher education is a formidable responsibility that challenges
even the most experienced technical project managers.
Data source dependencies, uncertain data quality,
changing information requirements, and urgency for
actionable information are but a few examples among the
multitude of challenges. What's more, the many kinds of BI
projects, ranging from strategic to operational, add to the
complexities. And then, of course, there are the various
technologies and skill sets involved in these projects--
ERP integration, data warehouse design, OLAP cube
design and system performance, data modeling, governance,
predictive analytics-- which compound the challenges.
The bottom line is that, currently, there simply is no
one-size-fits-all approach to BI project management.
Instead, BI project managers must learn how to choose
from among the traditional waterfall models, to the more
adaptive rapid prototyping methodologies, as well as other
project management methodologies and tactics that fit the
organizational culture, match the current skill sets of the BI
staff, and are appropriate for the project size and scope.
More importantly, they will probably need to move back
and forth among various approaches during the project
lifecycle-- say, across project planning, execution, monitoring,
scope control, completion, and postproject
maintenance and enhancements.
What's more, BI project managers often
struggle with a variety of enterprise organizational
issues that chronically inhibit success.
Some of these issues are technical; many are
not. Always at the core are the cultural and people
challenges. While not unique to BI projects,
issues of culture and people do seem to be particularly
problematic due to the variety of individuals
who necessarily are involved. Certainly,
business analysts, end users, technologists,
and decision-makers all must have a voice in
these projects if they are going to be successful.
Yet, with these diverse roles and those who
hold them, the challenges are many. Struggles
with a changing way of doing business, unclear
project roles and responsibilities, lack of agreement
on key decisions, finger-pointing over findings--
all of these challenges can lead to an
unpredictable project outcome.
Oklahoma City Community College faced
many of these challenges as administrators and
technologists there embarked upon a journey
into enterprise business intelligence. In little
over a year, however, the college has gone from
a mostly manual and ad hoc reporting system, to the development and deployment of a data warehouse
and online analytical reporting system delivered through an
internal portal the college calls the Institutional Intelligence
(I2) portal. Importantly, to reach this milestone, the college
laid out several focused projects, each designed to move
the institution farther along the BI continuum. Along the
way, both project manager and project team members
faced hurdles, but they adjusted and learned to focus on
outcomes and results instead of rules and hierarchical
roles. Here's how they did it.
Develop a Vision
First, OCCC spent time formulating a vision for a BI program
that would be ongoing and would contribute to developing a
"culture of evidence." This overall "umbrella" program, I2,
served as the framework from which all BI projects would
stem. The program called for improved information quality,
reduced manual information compilation and distribution,
and decision-making through data. Specific goals and objectives
were developed that were then used as guidance for
subsequent projects. Laying this framework helped in discussions
of priority with institutional leadership, and led to a
clear roadmap of projects as the program progressed.
A common mistake is to wait until the project is nearly
complete before unveiling it to end users. Employ rapid
prototyping so that the working product can be adapted
over time throughout the project, and not simply at the end.
OCCC's project leaders abandoned the traditional higher
education team approach that involves a committee of people
from across the organization with an unclear stake in the outcome
of such an initiative. Instead, to form the core I2 project
team, they built a cross-functional team of only seven professionals
from across the organization: two from information
technology, three from institutional research, and two who
represented key stakeholders. Collectively, the team had an
intricate knowledge of the goals, data challenges, and information
needs of the end users. Given that neither the project
managers nor project team members would be released from
their everyday roles and responsibilities, the team needed to
become "self-organizing." In other words, team members
needed to assign tasks to each other, coordinate and review
each other's work artifacts, collaborate on project activities,
make project-related decisions (together), and take on another
team member's tasks when necessary.
While the project managers still had a role in guiding the
team, instead of being "task masters" it was more important
for the project managers to run interference on behalf of
the team, when issues were difficult to resolve or political in
nature. Both the team setup and approach were new to
OCCC, and as one might expect, the changes were not
without glitches at the start. Over time, however, the structure
resulted in increased ownership and trust among the
The I2 team has since evolved to have a high degree of
credibility within the institution, and it conducts regular
meetings with representatives from across each of the business
units to communicate project status, gain feedback on
work products, and gather input on future BI initiatives.
One common mistake many project managers make is that
they wait until the project is nearly complete before unveiling
it to stakeholders. This is particularly problematic for
enterprise BI projects, as managing data and information
across the enterprise is more difficult and takes more effort,
coordination, and resources than delivering silo shadow
systems and point solutions. OCCC combated this by
employing a system of rapid prototyping so that the working
product could be reviewed and modified over time
throughout the project, and not simply at the end. This technique
helped to prevent projects from slipping or falling out
of scope, as it provided more time to make changes based
on user feedback. Certainly, there were setbacks, but rationalizing
redundant data and inconsistent business rules in
public at the end of the project would absolutely prove
embarrassing, and rapid prototyping was one way to minimize
the chances of building the wrong solution.
Get the End User Involved
A prerequisite for the rapid prototyping approach is the participation
of intended end users on critical project activities.
BI projects require much more involvement by end users
than do most other projects. Traditionally, stakeholders
participate in requirements-gathering interviews, project
reviews, and user-acceptance testing. Other than that, the
technical developers do all the development work with no
involvement from the users. But, besides creating an "us
versus them" atmosphere, this limited degree of involvement
forces enterprise BI developers to make assumptions
that often lead to unsatisfactory results. This is bad enough
on projects with well-defined scopes and deliverables, but on ill-defined enterprise BI projects, where scopes and
deliverables are often a moving target, it can be catastrophic.
Just think of how many times we hear that business
intelligence projects are late, over budget, too costly,
too complicated, and that the deliverables don't meet end
users' expectations and are not utilized.
At OCCC, throughout the life of each BI project, the
core project team regularly involved stakeholders from
across the institution. From the start, key end users who
might be impacted by the project deliverables were identified
from each area of the college. Approximately 15 to 20
people formed this larger BI review team and were briefed
regularly on the project status. When critical enterprise
issues needed to be resolved, this review team served as a
fount of business knowledge and a sounding board for the
core project team. Most importantly, though, the review
team members were provided with structured walkthroughs
and access to the evolving prototypes, to allow
them to provide direct feedback and validate the data they
were seeing. This not only helped to improve the quality of
the final deliverable, but also increased the ownership
stake of the actual end users themselves.
Plan-- and Be Nimble
In the end, a healthy and sustainable BI initiative (i.e., data
warehouse, data marts, cubes, reports, dashboards, etc.)
doesn't just happen; it requires careful planning from the
outset. Yet, most importantly, it is how projects are defined
and managed that will have a significant impact on the
initiative's overall success. Specifically, successful BI
project management is about flexibility, skill in interweaving
methodologies, and actively engaging stakeholders.
Attempting to use a single methodology simply will not
work. The traditional linear "waterfall" and "big bang"
methodologies, with their rigid order and highly structured
teams, are just not agile enough or fast enough to meet the
evolving information needs of today's decision-makers.
OCCC recognized the need to be more nimble as an organization,
in order to maximize its investment in enterprise
Clearly, an approach that incorporates a focused, self-organizing
team; rapid prototyping of work products; and a
high degree of end user participation throughout, will likely
yield more rapid results for your institution while at the
same time, increase ownership and trust in the output.