Root-Cause Solution for Problems with Data Quality & Silos: Concept Models
Extracted from Business Knowledge Blueprints: Enabling Your Data to Speak the Language of the Business, by Ronald G. Ross, 2020.
These days, every organization of any size tussles day-in and day-out with migrating, correcting, consolidating, and merging data. Huge amounts of costly resources are devoted to the problem. Here's what a seasoned executive recently said about IT in banking, where the problem is particularly acute:
"Up to 80% of all IT activity in banking is absorbed in supporting data integration and cross-application aggregate reporting."
Does all this activity actually resolve any of the underlying problems? No. It's like running in place on a treadmill going faster and faster. Stepping off, of course, is not an option.
If the data were of high quality, that would be one thing. But generally, it's not. Far from it. Organizations have simply been looking at the problem of data quality in fundamentally the wrong way. Time for a fresh perspective.
How did organizations get themselves in the current situation? Basically, a long-term application-centric approach is to blame. Each project often develops its own files and databases. The net result is local, narrow views that must be spliced together through external interfaces. The big picture isn't simply lost — it never existed in the first place. And without a concept model, one that is integrated and holistic, there can be no effective remediation.
What about reengineering business value chains? The heart of the matter is very down-to-earth. Integrating silos will never happen so long as workers in different parts of the value chain remain in different semantic silos.
The problem is not just one of functional silos. People (and machines) need to mean the same things by the words they use.
Unfortunately, there is no silver bullet for these problems. We simply need a better approach. The solution is a concept model.
How exactly does a concept model relate to data? A concept model is not some abstract model of data; instead, it is a concrete model for a different domain, the design of business knowledge.
Still, it's important for data for the following reason. You're simply never going to solve the problems of cross-organizational data integration and aggregated reporting using the same techniques that got you into the trouble in the first place. You need a concept model, a whole new foundation for business knowledge.
# # #