Where Data Warehouses Fall Short

The data warehouse seems like a perfect BI solution. It provides a central repository for storing all of an organization’s data that, in turn, gives reporting tools a single location where those tools can find and extract useful data that can then be analyzed and consumed.

The disappointing reality.

The reality is different. Picture a conventional warehouse with inventory not just coming in and going out via the front doors, but through windows in the walls and holes in the ceiling. Plus, inventory has been stored outside the warehouse at various locations, and it’s constantly being turned over. Trying to get a handle on all of that inventory at any given time would be incredibly difficult and probably impossible. In the typical data warehouse, a similar situation exists.

It is enormously time consuming to connect one data provider to the data warehouse, and most organizations will likely have a large number of data providers. By the time organizations standardize the reporting structure of a data provider, often the data structure or reporting requirements will have changed.  Using traditional ETL processes then holds important business questions hostage to a data manipulation process.  In effect, change happens quicker than results.

What to do.

All is not lost. Rather than seeking the ideal—which is unattainable—effective BI users accommodate to their situation. Flexibility is essential. For instance, there will always be a particular sales channel that is outside the normal reporting model, or a rogue spreadsheet that contains valuable and ever-changing operational data.

Bottom line: acknowledge that your data warehouse will never be complete. Yes, never. The data warehouse is very valuable—it is an effective hub for major software systems—but it is important to recognize its limitations. Rather than trying to eliminate exceptions, you would do better to accommodate them. If you do not, you will always have incomplete data and therefore decisions based on incomplete information.

Needed are BI tools that easily (ease of use is the key) expand beyond the core data warehouse and allow end-users to integrate a myriad of data sources into the BI conversation without requiring timely ETL conversions and integrations.

For more details about the problem as well as the solution, check out the full whitepaper, here.