Six myths of enterprise business intelligence

21 Jan 2015

How much of what we think or 'know' is untested belief? What if the opinion that we never question and unthinkingly repeat turns out to be patently wrong, making us seem naive?

Robert Fleming

A few examples:

The Great Wall of China is not the only manmade structure that is visible from space. Using the edge of space (100km above Earth's surface) as the vantage point, one can clearly see cities and even very large buildings in good viewing conditions.

Vikings did not wear horned helmets. It is a fiction created by Wagner in his opera The Ring Cycle.

But so what? We're all guilty, right? Popular myths are surprisingly common. But as they say, 'in the information age, ignorance is a choice'. We can no longer uncritically repeat embarrassing fictions with such blind confidence.

BI truisms that need revisiting

Myths and misconceptions are common in business intelligence (BI) too. Essentially, they're hangovers from a time when BI executed slow disk-based operations using structured, hierarchical query logic, and used all sorts of workarounds and compromises (cubes, data warehousing and reports) to deliver imperfectly on the promise of information at your fingertips.

Today, all the BI action happens in user-driven 'data discovery', not IT-driven report compilation. In-memory calculations and associative query logic deliver lightning-quick answers, with the added bonus of a tiny BI footprint, so queries can be done on the fly on mobile devices. Tools are easy to use and visual, balanced with governance (data integrity and security) to satisfy IT's need for control.

In the new world of data discovery - BI that anyone can do, we no longer need to believe and evangelise the following outdated platitudes:

Only decision-making managers need business intelligence: The hierarchical architectures of older BI stems from a hierarchical view of the enterprise. Today, people at every level of the organisation have access to information and decision making, increasing the need for a wider BI footprint.

Good reporting = good BI: Reporting is static information delivered to a large number of people or an organisation's top echelons. It is impossible to 'interrogate', which is what good BI does.

Fast in-memory BI will fix the adoption problem: In-memory BI is important for sure. But speed alone won't fix the problem of poor BI adoption if usability and desirability isn't addressed as well. Make it fast, easy and flexible, and you just might be in business.

We don't have the analytical skills: Why pay analysts or data scientists to interpret data? Humans have evolved natural analytical capabilities, including pattern recognition (distinguishing between clumps and single dots), outlier detection (noticing something different about a room) and categorisation (relevance detection). Tools should adapt to leverage this.

We need more visuals to help people 'get' data: Yes, visualisations merit a lot of attention by vendors, but a picture alone is not sufficient. Some tools have gorgeous visualisations but don't allow drilling down into data sets, and obviously that is anathema with today's gadget-empowered, internet-informed, application-savvy employee-consumers.

Better access to data = better decision making: Actually, having all the information in the world at their disposal did not help bankers to avert the sub-prime mortgage catastrophe and 2009 financial crash. One needs to be able to use data to gain insight.

New rules, new tools

Visionary author Malcolm Gladwell has said it takes around 10,000 hours of practice to attain mastery. The goal of BI vendors should be to take that out of the equation with tools that enable quick mastery.

That age is already upon us with business discovery platforms that deliver timeous, high-value, intuitive analytics for the many. The age-old ideas of business intelligence simply don't cut it anymore.