Teradata Magazine | Tech2Tech: Insider's Warehouse: Owning Business Intelligence

Many business intelligence (BI) programs have delivered a solid initial rollout, but program managers find challenges in getting beyond those first targets. Often the methods that ushered in the early ROI plan are too free-flowing and unmanageable when users are counting on the data warehouse for production needs. And once the concept is proven, recouping the investment requires a rollout beyond the initial set of core users. However, prospective users of BI require more than a well-built infrastructure in order to convert them to daily users. Attracting users—which is required for success—can be a difficult and elongated process.

In order to attract new users, it is vitally important to relate to your user community. Users come to the BI table from a variety of perspectives. Often users are completely unfamiliar with BI and may view it as an oxymoron. Those who are somewhat familiar may prove to be an even greater challenge, since they come with a preconceived standard that may be too low or too high for your current delivery capabilities. Both perspectives need to be managed through exhaustive communication and education.

When it comes to IT solutions to business problems, many prospective users have "heard it all before," and BI is likely just the latest in a long string of remedies. Users are looking for a fuller understanding of their business than has been provided before, and they are looking for your ability to fill the gaps from their previous experiences. They need reassurances that teammates from their side of the fence will be participating in the implementation process, that they will receive adequate training and that the BI solution is not just going to be dropped in production and unsupported. You need to reassure them that you are serious and you understand the full life cycle implications inherent in asking them to rely on it.

Finally, prospective users are going to be looking for the "deeper meaning" that comes with automating some of the work they do. This can cause unease, to say the least. In order to minimize this tension, it is very helpful to have tangible executive support for the program.

So, using this scenario as the basis from which prospective users come, I will go into some of the programs that can be put in place to calm the unease and fears and strengthen the relationship between the builders of BI and the users.

Activate the users
Most users are not co-conspirators in the success of BI. They are merely users. Little is done to "activate" them as co-owners who are vested in the success of business intelligence. Yet we have clearly seen that those programs that include strategies to activate users have done so with great success. The plain fact is users have roles and responsibilities in BI. Most will embrace those roles when they are made aware of them. But if the BI leadership team has not expressed what those roles are, users will not be asking.

To ensure usability, it is important that users undergo training before they become users. Sophisticated programs provide different levels of this training, which I will outline later in this article, but the key is that all users receive the right level of training for their needs. Providing training appropriate to the users has an additional benefit: By proactively addressing through training the common, mundane questions that everybody has, you'll reduce the number of individual inquiries from your users.

Users also have a responsibility to browse the metadata that you create. Think of it as the "manual" for the data warehouse. That's a high level of responsibility for the metadata. Investments in metadata—when made for the user's benefit—usually pay off.

Finally, we want to ensure feedback from the users—both good and bad. No feedback at all is the worst-case scenario. We need to build relationships with users, and users need to understand the evolving nature of the data warehouse.

Without understanding and acceptance of the evolving data warehouse, users are apt to feel that shortcomings are permanent. They're not—they're normal.

Data steward responsibilities

	Arbitrating the transformation rules. This is not an IT responsibility. Users will utilize the data and much of the data can be represented in multiple ways. User representatives should determine the ways in which this will happen.
	Determining the data population method. Sometimes data can be sourced from multiple locations. Data stewards make the sourcing decisions.
	Contributing business metadata. The business metadata is going to be most helpful to business peers. It should come from the business perspective.
	Supporting users. Stewards should understand the data intimately and support users who are getting acquainted with the data, with ongoing training as changes occur.
	Participating in program governance. Data stewards should contribute to the rollout plan for the data warehouse.
	Ensuring data quality. The build team should articulate the possibilities and make recommendations, but ultimately, responsibility for data quality lies with the user community, embodied in the data stewards.

Data stewardship
Data stewardship is all the rage again, although it is usually not implemented effectively. You will need a warm set of users to help tune the warehouse until it's ready for the masses beyond the high-touch zone. The stewards are those extended team members who happen to come from the business side. They, too, are users.

Your data stewards should come from the business subject area level (i.e., product, sales and customers) and should be management-level personnel in those business areas most impacted by each subject area. It's not a job title. For the core subject areas, this will be the person on the working team who is appointed by the executive sponsor. If there is a master data management strategy in place, the data stewards could very well be focused there. Regardless, the stewards are essential to getting your BI program out to the masses and to achieving greater levels of ROI.

Program governance
Program governance is the body that steers development activities. Representatives from data stewardship subject areas must be actively involved in these decisions. All programs should have a three-year plan in place, but the only way this will work is if the plan is revisited frequently and, if necessary, refined. In each iteration, it must be confirmed that the activities on the plan are still the best next targets for the efforts of the resource-constrained BI team.

More and more programs are targeting financial ROI with their BI programs, and rightly so. However, linking BI to ROI is, to say the least, a matter of interpretation. That interpretation looks self-serving and misplaced when coming from the build team. Program governance should direct that link.

Also, we know there are budget cycles in corporations. If BI is profitable, as it should be, it should not find itself on the short end when budgeting is needed for additional disk space, consulting help to get through difficult periods or to accelerate progress, or new data access tools.

User training
Training is one of the critical components of user care, yet it is a task that is usually uncomfortable for the build team. However, a little can go a long way toward increasing productive usage and reducing the overhead on the team for support through leverage. True training, to the level that is effective, requires material preparation, room reservation, lab development and schedule management, among other things. You will want to do all of this if you indeed want to require the training before providing business users with access to the system.

Data stewards should deliver fully half of the training—the latter part. The training should last half a day to a whole day, and the curriculum should be refreshed when significant new content is developed.

While the temptation to have data access tool vendors conduct training may be strong, I would avoid this. Standard training is seldom appropriate for end users with short attention spans. Craft training around the topics most useful for your users and provide reference to other programs for ongoing or deeper interest.

The ultimate in user-training frequency is for some measure to occur as part of new employee orientation, followed by more detailed training that correlates to job responsibility and user profile to occur at a later point.

User support plan
We want to give our users an outlet for their ongoing questions and support needs. BI, as much as anything, needs a support line. Perhaps manned on rotation by the BI team, it provides the users a sense of comfort that help is a phone call or a mouse click away. Many programs use "message board" technology to field user questions. I expect many more will use blog technology in the future.

The advantage of this type of support is the leverage you gain with the answers. The answers will be kept online for posterity, preferably categorized and searchable. This is an important feature of the intranet Web page for BI, where training schedules, data models, iteration plans and contact information are kept for the user community.

Your user support plan should include a strong measure of business support as well as technical support. Data stewards should be prepared to field calls related to BI usage, at least during business hours. This indicates how strong the business involvement in BI needs to be.

Public relations
BI is a living, thriving entity within a business. As data is made available, initially only a small fraction of the eventual uses of it are enabled. Most of the time, they are not even part of corporate awareness. A public relations program can help uncover those uses and provide a strong data point to program governance, which will then determine the next steps. Common elements of an effective public relations program include:

	A branding name for the program
	Shared guiding principles within the BI team, in order to present a consistent picture to the user community
	A subject area orientation: Enterprise subject areas form nice break points for mentally dividing a program
	"Chalk talks": Find hot, interesting issues and drum up interest and support from current and prospective user populations with talks over lunch
	User e-mail lists: Some will issue newsletters, but lists can be used to issue alerts to the community about expected and unexpected outages, new additions of data or function and anything people would find interesting.

Executive sponsorship
The BI program must have high-level and sustainable sponsorship in order to succeed. The executive sponsor must be politically viable and able to garner and retain adequate resources for the construction and maintenance of the program.

The executive sponsor can be responsible for strategic discernment and prioritization over the major additions of usage, subject areas and data sources to the data warehouse. It is imperative to the success of the data warehouse that the executive sponsor take leadership and provide vision to program governance.

A little top-down support in the form of organizational direction setting goes a long way toward ensuring that efficiencies are gained through widespread adoption of BI standards.

Workload distribution
Workload distribution for the BI team that includes minor enhancements and fix-up work—as well as major enhancements—ensures that there are ongoing resources available to support the program. It is easy to focus on highly visible major enhancements, neglecting the support required for the development that is already in production, but to do so could put user acceptance at peril. To avoid jeopardizing the continued viability of the overall program, be sure to allocate appropriate team cycles to run or support the important programs.

Running a successful and user-friendly BI program is a lot like running an intelligent business. An intelligent business has a customer base that it treats with respect and courtesy. It supports them with many programs while it seeks to grow new business. Users are the lifeblood of BI. To be successful, BI needs to be run like a program, not a project, with numerous components that support and nurture its customers. T

William McKnight is the senior vice president of data warehousing for Conversion Services International (CSI), a leading provider of professional services focusing on strategic consulting, data warehousing, business intelligence and information technology management solutions. McKnight is a former award-winning IT executive in data warehousing. He contributes a monthly column to DM Review and frequently keynotes and speaks on the subjects of data warehousing and BI internationally. McKnight serves as a judge for industry awards, has been an expert witness and is an active data warehousing practitioner.