January 14, 2002

7 Habits of Highly Effective OLAP Development Teams

These strategies can help your development team converge skill, desire, and knowledge in a manner that promotes ROI

By Norman Comstock

In business, the need for quality analytics has never been higher. Managers are working hard to provide the foundation and tools for analysis that will lead to the knowledge that can change a business's course of action. This effort often involves the implementation of online analytic processing (OLAP).

Once a fringe technology, OLAP is now on the shortlist for custom solutions; it's embedded in packaged analytic applications, and it serves as an engine for CRM solutions. Despite its relative ubiquity, OLAP is still commonly misunderstood from both developer and user points of view. In many cases, this misunderstanding — which can jeopardize an application's business usefulness and ROI — is an outgrowth of the development and deployment process.

In his 1990 book The Seven Habits of Highly Effective People, Steven Covey identified seven principles for gauging and fortifying personal effectiveness. These principles may appear to reflect common sense; however, many people fail to internalize them. The same applies to developers of analytic applications: Common sense is often ignored during the development and deployment process.

OLAP technology by itself doesn't guarantee that users will take full advantage of analytics; rather, personal interpretation is an essential ingredient for meaningful analysis. Therefore, to ensure that OLAP technology is used effectively in your organization, you need to provide a "compass" for guidance.

Organizations on this quest will ultimately select a methodology and tools for the job. But that process is futile if the team responsible for delivering the solution lacks the necessary habits.

Covey defined a habit as the intersection of skill, desire, and knowledge. Skill is the ability to execute how to do. Desire is the motivation, the sense of urgency to want to do. Knowledge is the understanding of what to do and why. All three elements are necessary to foster the habit.

Consider the hypothetical scenario faced by a successful, fast-growing catalog clothier. The vice president of sales is on a mission to understand a sales slump that has developed in several new regions over the past two months, the effect of pricing changes for the quarter, preferences of top customers, effectiveness of the marketing campaigns, and so on. Thus, three sales analysts have the seemingly never-ending challenge of compiling the tremendous volume of sales data and answering myriad questions that arise out of summarized reports from accounting that suggest wild variances in actual, planned, and forecasted sales. The analysts are frustrated because they're unable to readily identify whether the variances are due to specific products, regions, price points, sales mix, and so on. The VP of sales realizes the need for a more efficient process and subsequently sponsors a project to build an effective sales analysis application.

The development team comprises two technically savvy developers who had previously supported the company's online storefront. Giving the assignment, the project manager, who has several projects running simultaneously, neglects to articulate the pain the sales analysts face and also neglects to involve the sales analysts in the initial requirements discussion.

With little business input and an accelerated deliverable date, the development team assumes that most users of the application would only be interested in summarized information. When the application is delivered, they quickly learn that the application meets few requirements for the sales analysts, the product managers, the region managers, or the VP of sales.

What happened? The development team was technically competent; they had the right technical skills. The development team was committed to produce the deliverable in the designated timeframe; they had the desire. But the development team didn't fully understand the sales application's important ability to streamline the analysis process to help users gain quicker insight into the details behind the summaries. They assumed they knew what to do, but they never investigated why the sales analysis application was important. Skill, desire, and knowledge failed to converge.

HABIT 1: ENVISION THE SOLUTION

The leaders of the development team must have a shared vision that is clearly communicated to all constituents. The constituency consists of the stakeholders at the senior levels of the organization responsible for executing the strategies of the business, the line of business managers executing the tactics that enable the strategies, the business analysts monitoring the daily activities and forecast, the solution's developers, and those who will maintain and amend the application to adapt to changing business needs. In other words, everyone has to understand the solution's business goal: to provide a framework for business users to intuitively discover and analyze meaningful, clean data in a manner appropriate for each individual.

This tenet of OLAP was originally espoused by E.F. Codd; he believed that end users should concentrate on analysis rather than collection in order to synthesize more data in shortened cycles. The goal is to draw conclusions that managers can use to make proactive decisions rather than reactive decisions. To do so, they need to understand the broader data context in order to pinpoint the root cause, rather than prescribe modifications that only address symptoms.

Let's return to the hypothetical scenario faced by the catalog clothier. The VP of sales uses the newly developed sales analysis application with the dimensions customer, product, region, and time, to conclude that sweater sales are influencing the sales slump and price reductions are necessary to stimulate sales. If a promotion dimension had been available, the conclusion might be that no promotions have involved the sweater product line in the new sales regions.

True analysis is context sensitive. If all you're doing is providing a black box to crunch numbers and generate derived metrics, someone may not see the parameters influencing a positive or negative trend. The trend may be subtle over time or influenced by geopolitical and economic conditions, changing culture, or demographic shifts.

Effectively envisioning the solution on the front end, however, forces you to consider other important elements. These other principles will guide you toward the vision.

HABIT 2: DESIGN WITH THE END IN MIND

If you are practicing Habit 1, your focus may naturally move toward questions regarding the application's look and feel. This focus is productive. But to be effective, you should review the importance of perspective. The various and sundry people involved in the OLAP process will doubtless have different perspectives: the power analyst may want full drill-down to the atomic level, sales managers may only want to consider the summarized subset of their product or their regional responsibility, executives may need the bird's-eye view to confirm that the ship is on course.

The architecture may look drastically different if an important constituent isn't addressed. For example, more dimensions, more levels for the dimensions, or alternate hierarchies might be necessary. Therefore, it's crucial that you get input from all the stakeholders and potential users involved; the presentation layer should be sufficiently flexible to cater to individuals.

Remember, the choice of OLAP engine and storage method — be it multidimensional, relational, or hybrid OLAP — is largely dictated by the needs of the stakeholders. By designing with the end result in mind from the outset, you can balance the various requirements involved to create a back-end solution that supports a personalized interface. For example, the number of dimensions needed for comprehensive analysis may necessitate an OLAP engine that gracefully scales the number of dimensions to satisfy diverse improvised analysis needs, faster query response, or both.

HABIT 3: REMEMBER FIRST THINGS FIRST

Let's assume that you're on track with the first two habits. Now is the moment of truth: What do you tackle first? Too often, the easy problems are tackled first, the budget disappears, and the more difficult and meaningful problems are left unresolved. You can avoid this trap by itemizing the analytic issues into distinct quadrants, which will provide a management matrix useful for long-term planning. (See Figure 1.)

If you have taken the decision-support system approach to understand the importance of the information and the manner in which it is presented, such as dynamic or static tabular display, business chart, or geospatial visualization, you may already be on track to committing resources (time, people, and budget) to those things that matter most. Can you do this alone? Yes, but your views may not align with those of all your constituencies. Thus, take a survey of everyone's priorities, and then have the constituency groups organize them into the quadrants. Although you may not have any issues fall into quadrant IV — not important, not urgent — not every group will agree entirely on the issues to tackle as I, II, and III. Look for the union of the constituencies — the consensus. Publish the results, and if need be, have the project sponsor determine the final priorities.

• Privacy
• Your California Privacy Rights
• Copyright © 2008 United Business Media LLC