Self-service business intelligence (BI) has been the holy grail for BI professionals for a long time. Yet almost two-thirds of BI professionals (64%) rate the success of their self-service initiatives “average” or lower. Newcomers to BI struggle even more, with more than half (52%) rating their attempts at self-service BI “fair” or “poor.”
One reason for these less-than-stellar numbers is this: Implementing self-service BI is more complex than it looks. It’s not a one-size-fits-all program. BI users come in many different shapes and sizes, each with unique information requirements. This report lays out several frameworks that explain how users interact with information and then maps elements of each to BI functionality and categories of BI tools. This mapping is critical to success with self-service BI.
At the highest level, there are two types of self-service BI: There is report interactivity geared to casual users and report authoring and analysis tools geared to power users. Giving both types of users the same tool is a recipe for disaster: Either casual users find the tools too complex or power users find them inflexible and limiting. A better strategy is to deliver the right tools to the right users based on their information requirements.
The 80/20 rule provides suitable guidance for meeting these needs. It says that 80% of the time, casual users need interactive reports and dashboards, while 20% of the time they need ad hoc tools to create their own reports and views. The reverse is true for power users. The primary challenge with self-service BI is meeting the 20% ad hoc requirements of casual users.
Rather than provide casual users ad hoc authoring tools that they won’t use, it’s better to cultivate superusers in each department to support the ad hoc reporting requirements. Superusers are businesspeople who gravitate to BI reporting tools when they are first deployed and become the go-to people in their departments for obtaining ad hoc views of data. Superusers are the primary targets for self-service authoring tools, and they are the key to achieving success with self-service BI in most organizations. But to take advantage of superusers, corporate BI teams need to provide them with ample training and support as well as a solid data environment that contains consistent definitions of key data elements built into a robust semantic layer.
Although casual users need tailored views of information, power users need the opposite. They need to use a variety of tools to explore and analyze data as well as publish sanitized views to casual users. They also need to be made full-fledged members of the corporate BI environment. Left to their own devices, they’ll create spreadmarts and data shadow systems. One way to corral power users without undermining their freedom is to create data sandboxes inside the analytical ecosystem, where they can blend local and corporate data to their hearts’ content.
Another key to succeeding with self-service BI is a hierarchy of functionality for the two types of self-service. For instance, users who consume information, rather than produce it, require different levels of data interactivity, ranging from viewing static data to navigating that data to eventually modifying, exploring or modeling that data. A BI manager should create a hierarchy for information consumers and producers and then map categories of business users to each level in that hierarchy.
To make efficient use of the hierarchy, the BI manager should select self-service BI tools that offer functionality on demand as business users are ready and willing to use the new capabilities. Good self-service BI tools either hide these functions behind an unobtrusive but visible icon or allow administrators to enable them. Regardless, it’s imperative that BI managers continually monitor user abilities and requirements, which change rapidly, and adapt tools to align with user needs.
Other salient facts from the research include the following:
To succeed with self-service, BI managers need a deep understanding of their business users and capabilities of their BI tools, and then they have to map requirements to tools. They need to continually monitor the capabilities of business users and BI tools because both evolve quickly, and that can quickly make a well-crafted self-service strategy obsolete.
- BI self-service gets barely passing grades: 64% of survey respondents gave their self-service BI initiatives a grade of “average” or worse, while 31% rated it “good” and 5% “excellent.” BI beginners were more apt to give self-service BI lower grades, while advanced BI users gave it higher grades.
- The biggest challenge to self-service BI is counterintuitive: 73% of BI professionals said it “requires more training than expected.” That’s because most casual users find it difficult to use self-service BI tools, and superusers need lots of support to become proficient BI developers.
- BI penetration is minimal but growing: Today only 26% of employees use BI tools, but this is an improvement from 2005 when the penetration of BI tools was 18.5%. Another 37% use the output of BI tools today, while 42% don’t use BI tools at all.
- BI usage today is still rudimentary: 47% of business users just view static reports or dashboards, and 29% just navigate predefined drill paths. Twelve percent modify BI reports, 9% explore pre-existing data and 6% model it. Newer, more analytical BI tools promise to accelerate usage.
- Among information producers, the most common way to create new reports and dashboards is to craft them using a semantic layer (35%) followed by assembling them from pre-existing report parts (20%).
- Traditional ways of delivering self-service BI dominate: The most commonly used self-service BI tools are BI tools with semantic layers (69%), followed by desktop analysis tools (63%). Up-and-coming self-service BI tools are in-memory visual discovery tools (25%) and BI mashup tools (21%) that support both power and casual users.
Read the entire study.