Self-service BI isn't just for the "average" end user with limited technical and analytical expertise, says Forrester analyst Boris Evelson. Business analysts need to do predictive analytics, multidimensional querying and data mining. Knowledge workers and power users want to do ad hoc querying and generate their own reports and views. Self-service BI platforms enable them to do that while shielding them from the underlying data infrastructure, so they don't have to keep asking IT for help.
Flexibility was key at IXI Corp., a unit of Atlanta-based Equifax Inc. that provides risk and performance management consulting services. With IXI's old BI system, it was too difficult to make any changes to a data report, says Russ Ayres, the company's senior vice president of customer insight. Requests for changes meant that the underlying data models had to be revised and then approved, which was a slow process. Hard-coded data structures weren't cutting it with IXI's customers, whose data needs change on a daily basis, Ayres explains.
The company addressed these challenges by using Tibco Software Inc.'s Spotfire. IXI analysts use Spotfire to do rapid and flexible data querying across multiple data sources, Ayres says. On average, creating a new view takes a quarter of the time it used to, he adds, "so we're about four times as productive."
Despite their enthusiasm for self-service BI, IT executives acknowledge that easy-to-use BI tools can be dangerous because of the power they put in the hands of end users.
Data governance, security, and centralized monitoring and control of user interactions are critical for any BI system, but particularly for self-service setups that give less-technical end users direct access to the corporate data infrastructure.
"Governance is where we [IT professionals] come in," says IXI's Ayres. "When you give someone a loaded weapon, they can always shoot themselves. Someone could do a broad search across a 5TB database and bring a server to its knees, or worse. BI doesn't stop you from making the wrong choices; it just helps you make them faster."
To avoid this problem, Ayres' team has built a layer between the user and BI tools, "so you can dance around the playground, but within limits." For example, an employee might be allowed to run metrics in a data mart at some levels but not others, and he wouldn't be allowed to summarize across different levels.
Most of the major BI platforms support role-based access control, through Microsoft Active Directory or any other LDAP-compliant global directory. Packages offer different degrees of granularity: For example, a system might be set up so that a particular user group can access only a subset of data, or even just specific data fields.