Data Warehouses Disappoint
We meet customers every day who wonder why their warehouse solutions have failed to live up to data analysis expectations. The reasons are fundamental.Warehouses are inflexible by nature. If there are a million things in a storage bin, it's difficult and time-consuming to move or re-classify them. Unfortunately, data analysts are always asking, "What if these things were in that bin over there? What if we divided this area into fourteen sub-categories? What if we rolled up this way, instead of that way?"
Analysts always want to invent new organizations for data, try them out, discard them, and then try others. There is no such thing as a hierarchy that never changes, and nothing worse than a design-by-committee schema that's useless for ad hoc analysis.
With data warehouses, change is an onerous process. IT wizards or other specially-trained people have to get involved. Sometimes only the warehouse vendor can make changes (this is common with spend analysis products, many of which are packaged with data or consulting services). The process can take days, or sometimes weeks. The cost and inconvenience of change means that the data warehouse doesn't change very often, and it certainly can't change to support the random analysis of the day.
Warehouses are inflexible by definition. Let's stipulate for a moment that changes to a data warehouse could somehow be made easily, quickly and cheaply. This is untrue of any data warehouse we've ever seen, but let's suspend disbelief for a moment.
Can we then support the data analyst who wishes to group Puerto Rico with Europe, cut the Organization structure by product line instead of cost center, and derive a new Contract dimension as a function of Commodity and Vendor? Of course not, because the other 100 people who are sharing the data warehouse will be very unhappy when it changes out from underneath them. In fact, in order to change the structure of the warehouse, we'll first have to get those 100 people in a room (or at least the 10 decision-makers to whom they report), and we'll have to get everyone to agree on the changes.
So, the very idea of a "data warehouse" that supports ad hoc reporting is flawed, because the warehouse can't change by definition. Therefore, the only analyses that can be performed are those that fit neatly into the pre-defined structure of the data warehouse — a very small slice indeed.
Since the analyst can't use the data warehouse for ad hoc reporting, what does he do? Exactly what he's always done. He extracts raw transactions to Microsoft Excel or Access, and then he fools around for a few hours or days until he's got a properly-organized dataset. Then he spends more hours or days building the report.
This means ad hoc reporting is slow and expensive. There may be important insights buried in the data, but you'll never realize them, because there aren't enough hours in the day or analysts in the company to explore the dataset properly.
With BIQ, each analyst has his own private playground. Data organizations can be changed instantly. New dimensions can be derived trivially. Complex data mappings can be performed easily and quickly. Then, the full power of BIQ's advanced OLAP and data display facilities can be utilized to generate reports quickly -- and to view the same report from hundreds of different perspectives, with just a few mouse clicks.
BIQ works alongside your data warehouse to provide the capability you thought you were buying originally — at very low incremental cost.