Without effective, governed self-service data preparation and data discovery, information becomes noise, trust in the information is diminished, and effective collaboration becomes much more difficult. Everything is louder than everything else.
Self-service business intelligence platforms provide significant benefits, but have also contributed to a new trend: the "wild west" of proliferating BI silos, inconsistent business definitions, no data lineage, and no single version of the truth.
"Spreadsheet hell" has been replaced with "Self-Service BI hell." As Boris Evelson (of Forrester Research) recently told me via email, “We increasingly hear from our clients that BI silos are now proliferating. Basically these platforms are now becoming the new spreadsheets”.
That introduces risk. Consider a recent article in The Economist (“Excel errors and science papers”) which reported:
“…they had accidentally omitted five rows of their spreadsheet when calculating an average. When included, the missing figures weakened the paper’s conclusion substantially.”
Self-service is all about speed and agility, allowing business users to follow their own intuition and answer their own questions rather relying on IT. In the 1990s, we used to call it the “next question dilemma”: It’s impossible to predict the next question a business user is going to ask until they’ve seen the answer to their previous question. Self-service data discovery needs to be iterative, exploratory.
But can the "need for speed" in business decision-making be reconciled with the need for governance? According to Howard Dresner, governance of BI content creation and sharing correlates strongly to success with BI, improving information consistency and accelerating group-based decision making.
BI governance includes such things as BI lineage, impact analysis, facilitating collaboration, and content reuse, reducing content duplication. In the BI industry, we’ve seen what Wayne Eckerson recently called a “pendulum swing” away from over-governed to un-governed BI. The pendulum is now swinging back, because business users are now starting to ask questions like:
· How do I trust the decision being made?
· How trustworthy is the data?
· How timely is the data?
· How do I communicate the decision, the thought process behind the decision, and the facts supporting the decision?
Added complexity arises from the increasing number of additional sources of information available to a business user.
I recently spoke to a customer in the financial services industry who said they receive data such as AML (anti-money laundering) data from external sources, usually in a flat-file format. Users need to merge or blend these data sources with internal data to produce their dashboards and reports, supporting their business decision-making. Due to the time-sensitivity of the data, the users needed more of a self-service approach, but still exercise some governance to retain confidence in the information.
In another example, a business user at a government customer used to complain the BI content they received had no “context.” What were they looking at? What does this number mean? How was it defined? When was it updated? What is it filtered on?
It continues to surprise me after 25 years in the BI industry, that most BI output still doesn’t contain this kind of basic contextual information.
This is perhaps why many business meetings are still dominated by debates about the numbers and whose “version” of the numbers is correct, instead of actually making productive, collaborating business decisions. Without the ability to properly parse data and give it context, BI can create more noise than signal. Deep Purple might approve, but it doesn’t do businesses much good.
(About the author: Patrick Spedding is managing director of business intelligence research and development at Rocket Software. He is also an evangelist and strategic advisor in the areas of business intelligence, analytics, and strategy and risk management).