I hear a lot of reasons people avoid a data warehouse: It takes too long to build, data warehouse projects are expensive and often end in failure, I’m good with a dedicated table to support each of my reports.
In future posts, we’ll look more closely at some of these justifications to avoid a data warehouse, but first here are my top three reasons why you SHOULD have a data warehouse:
- A data warehouse will make it easier to get to the answers you need, when you need them. Traditional databases store data in many small tables. Each of these tables will be joined to others.
Sometimes, it’s even possible to join groups of tables together in completely different ways and get significantly different answers to the same questions. A data warehouse simplifies the join paths, making the joins between tables much more intuitive. - The data warehouse is organized around groups of tables that each focus on ONE business process. By “sharing” some of those tables across multiple business processes, it becomes much easier to do analytics that cross over group and departmental borders. You can create meaningful views of Sales in one group of tables and Returns in another. Thanks to the elements in common between Sales and Returns, it becomes much easier to then create a unified look at both Sales AND Returns.
- Change. Traditional databases don’t often focus on tracking changes in the data. Imagine a scenario where you are watching sales on the East Coast. John Smith is a very big customer on the East Coast, until the day he moves to the West Coast. Would you want the history of all John’s purchases to now be credited to a sales manager on the West Coast? Probably not.
Traditional databases might often overwrite John’s address, making it look like all his purchases were initiated on the West Coast. A data warehouse can tell the difference between the purchases John made on the East Coast and those he made on the West Coast.
These are MY top three. What are YOUR favorite answers when you are asked, “Why build a data warehouse?”
