The pendulum of the Business Intelligence development paradigm swings between two poles in opposition, based on the degree of autonomy of system users to answer their own questions. This is especially true in situations that exceed the original set of parameters for which the underlying Business Intelligence system was designed to address. Here we examine the problems behind self-serve BI.

Initially, data analysis systems were exclusively the province of highly-trained data experts (often DBAs) who knew the content and structure of data and the storage systems exhaustively. And the company’s data users were expected to make formal requests for data from them. The operators would then write a query, run it and retrieve the data, prepare and format it, and return it. Analysts would review the results, formulate new hypotheses, and request more data, or even the same data aggregated differently. This process could repeat over and over as each business question was refined, answered, and opened new questions.

While this approach was the most sustainable on systems because the IT operators were acutely aware of how their queries ran and how they could be tuned to perform with as little impact as necessary, it nonetheless had the inescapable disadvantage of making the analysts’ business intelligence process (though it wasn’t called that at the time) incredibly frustrating and cumbersome. Business users clamoured for more freedom to get the answers they needed out of their own data.

In response, vendors developed tools that permitted companies to deploy intuitive-yet-powerful data interfaces, with the intention that non-technical users with a minimum of training would be able to develop their own reports to answer their questions without having to consult with IT. These tools gave users, who knew the business but not necessarily the data, the ability to ask incisive questions quickly. Unfortunately for the IT teams responsible for maintaining and delivering service through these BI platforms, those users often had little or no familiarity with the technical manner in which their data was being retrieved for them. An overzealous user with a few mouse clicks could easily create and submit a query that might run for hours or cause a dependent system to slow down or fail.

This method, sometimes referred to as “self-serve BI” or “roll-your-own reporting,” split the vendors’ two audiences: business users loved it for the freedom and autonomy it created to answer questions quickly and iteratively, while IT organisations who had to make it work hated the impact on over-stressed systems. They found unrestrained use consumed limited computational resources disproportionately.

Both groups had legitimate concerns — the pace of business was accelerating, and business users needed answers faster than ever before to exploit new openings in their markets, identify opportunities to lower costs, and advance against their competitors. Meanwhile IT had to support these systems, ensuring that they were robust, responsive, and extensible. More users demanding more data in more ways, faster, and more often, created an unnavigable maze of competing priorities amid apparently boundless demand for service.

Organisations and the industry as a whole answered both of these challenges with different modifications to the tools themselves and some of the overall implementation strategies, however, the pendulum of focus has remained on the side of the business users for over a decade, and only now is swinging back toward IT.

In the next article, we’ll explore ways in which self-serve BI is changing amid financial and technological challenges faced by companies right now.

DataHub Writer: Douglas R. briggs
“Mr. Briggs has been active in the fields of Data Warehousing and Business Intelligence for the entirety of his 17-year career. He was responsible for the early adoption and promulgation of BI at one of the world’s largest consumer product companies and developed their initial BI competency center. He has consulted with numerous other companies and is regard to effective BI practices. He holds a Master of Science degree in Computer Science from the University of Illinois at Urbana-Champaign and a Bachelor of Arts degree from Williams College (Mass).
View Linkedin Profile->
Other Articles by Douglas->