Business Intelligence clearly provides proven answers to known questions while out extensions, in this case, Data Discovery, it provides fast answers to new questions formulated by the business user. For example, when the business intelligence report says that warranty claims on the top-selling product went up 15% last month, the new questions are “What changed? What’s the root cause? What are customers saying about this? That exploration happens in a discovery app.
And the relationship goes both ways. Data Discovery creates new KPIs for the BI stack to deliver. For example, a consumer packaged goods company learned that preference for seemingly unrelated brands was highly correlated in certain customer segments. This came from a social media discovery app and suggested new KPIs they quickly pulled into their operational BI system.
Far from replacing their BI systems with data discovery, our customers have instead been able to get far MORE value out of their existing BI systems because they are able to re-focus them on solving the problems they are most effective for, and creating new practices around data discovery to get fast answers to new questions.
Just like NoSQL solutions solve different problems than relational databases, Data Discovery solves new problems that are different than traditional business intelligence and reporting:
The fact that data is available immediately creates demand for it. As more application, consumer, sensor, and mobile data is available to the business, the more the business wants to use that varied data for daily decisions that today get made on intuition and opinion.
In analytics, big variety is a bigger problem than big volume because it can’t be solved by more processing power alone. In addition, the cost and time required to combine diverse data together must come down.
The people making these decisions are experts in the business, not in writing SQL queries. They need a user experience that’s simple to learn and use and this is a core capability of Oracle Endeca Information Discovery.
It combines structured and unstructured data from inside or outside the company. An enterprise solution must work with the full range of data that matters to an enterprise, including multiple structured sources with diverse schemas, like the vehicle warehouse and quality touch point application data; including unstructured data like the long-form text descriptions in the warranty claims; regardless of whether the data is inside the company, like the warehouse, or outside the company like the NHTSA claims or JD Power data.
It delivers in-memory performance, but is not memory bound. An enterprise solution must maintain fully interactive query response times even when the data is too big to fit in memory. Endeca realized this years ago when it combined search and browsing in eCommerce because search indices are often too big to fit in memory. Oracle EID is written for multi-core, multi-processor servers and uses proprietary optimization algorithms to exploit the full memory hierarchy from on-CPU cache all the way down to disk. It is a solution for provisioning targeted discovery apps. It provides IT with a new capability to quickly deliver discovery apps wherever the business needs them.
Over time, we will see strong adoption of Data Discovery applications that further compliment and augment Business Intelligence solutions and why we will continue to see polyglot analytics take hold.