This article and others in the series are excerpts from a recent report on self-service analytics. They also form the basis of a half-day course that Wayne Eckerson teaches at TDWI conferences. 

Read - Part I - Self-Service Analytics: What Could Possibly Go Wrong?

Read - Part II: One Size Does Not Fit All: Customizing Self-Service Analytics For Business Users

The challenge with self-service analytics is giving business users the right amount of freedom to meet their own information needs. Too much freedom spawns data silos and distrust. Too little freedom creates backlogs and frustration, causing business users to circumvent IT entirely. (See “Self-Service Analytics: What Could Possibly Go Wrong?”)

What’s needed is a blend of self-service and governance that maps to the way both organizations and people work. There needs to be both iterative and bi-directional workflows to ensure that business users get the right information at the right time and organizations can maintain a clear, coherent data dialect that everybody understands and trusts. 

These workflows need to accommodate different user needs for information and access. Not all users need the same level of self-service. (See “One Size Does Not Fit All.”) In fact, the vast majority of users (~90%) don’t need self-service at all—or at least, not self-service in the traditional sense. What they need is “silver service”—highly tailored reports and dashboards populated with curated data—in other words, business intelligence (BI) served on a “silver platter.”

Contrary to what many vendors proselytize, the goal of BI is to minimize the need for self-service, not maximize its use. Most business users are not hired to crunch data and design reports and dashboards. If they do, it’s by necessity, not desire.

The goal of BI is to minimize the need for self-service, not maximize its use.

A well-designed BI environment with iterative and bi-directional workflows allows casual users to consume information quickly so they can do their jobs, not spend nights and weekends crafting custom reports. And it leaves the dirty work of preparing and designing reports to data scientists, data analysts, and IT professionals who are hired to do such work.(See figure 1.)

Figure 1. Self-Service Analytics Workflows

Left-to-Right Workflow. In most organizations, data flows from source to target, getting more refined and curated the closer it gets to business users. Traditionally, IT departments manage the curation process for both data and the creation process for reports. Consequently, IT often becomes a bottleneck that stands between business users and data.

With self-service analytic tools, however, power users can participate in the curation process. Using data preparation tools, they can transform, blend, and enrich enterprise and local data in an iterative manner. With visual discovery tools, they can visualize and analyze data and share their findings with departmental colleagues and managers. These casual users then view, search, modify, and discuss the power user reports and act on the results.

Right-to-Left Workflow. Equally important is a reverse workflow where business users feed requirements back into the curation and creation processes. After analyzing data in reports and dashboards, consumers better understand their requirements and can propose them to power users in their department. In turn, power users can build those requirements into an ad hoc report that casual users can review and refine.

If the report is popular, power users can submit the report to a governance committee to review. The report serves as a prototype for a new production report. The committee inspects the report’s embedded metrics and data definitions to ensure they align with corporate standards. And the IT department determines how to “productionize” the ad hoc report with adequate scalability, security, and reliability. (See figure 2.)

Figure 2. Report Workflows

Watermark. If the proposal is approved, the report then carries a seal or stamp of approval. This mark indicates to business users that the report is “safe” to use for decision making. The watermark helps users distinguish between curated and non-curated data. Before long, business users start refusing to use reports that don’t carry the seal of approval. This creates a culture of governance from the ground up. Rather than work independently, power users recognize the value of working through formal channels to publish reports for broad-based consumption.

Summary 

By understanding data and report workflows and the roles of different business users, organizations can accelerate the delivery of information to business users and eliminate BI bottlenecks. The result is a governed self-service environment which balances the business’s need for speed, agility, and freedom with an organization’s need for standards, control, and governance.

Read - Part IV. How Watermarks Can Transform Your BI Program