Why you ALWAYS need a specific database for business data analysis

Originally Posted on 10/15/2009 

It is often said and widely acknowledged that a datawarehouse for business data analysis is something for the big guys. I do not agree at all! Having little data does not mean that the environment is not complex. I had small customers which required very advanced analytics to keep up with what was going on. 
So, I try to explain as simply as I can why you need a database for business data analysis (i.e. a datawarehouse) EVEN THOUGH you are a small or medium company, and just have few data and your server that can handle all the workload etc.

1) Your history is there even if your OLTP systems blows up. It's not a matter of backups: the vendor goes belly up and there's no further support, your business changes too much to keep using the old system, your new CEO loves a different SW etc.

2) Building a datawarehouse forces you to think to an analysis model for your business. A stable model makes comparisons with the past possible. The need for changes in a data analysis database mirrors the need to change the company strategy; changing too often means that your company does not know where it is going.

3) Building such a database forces you to think to key performance indicators for your business. There are figures for every business; "magic numbers" that give you an immediate idea about what's going on. No business, no matter how small it is, can do without some key performance indicators. Often, they're not formally defined, but they do exist in people's minds.

4) Likely you have different systems with different master data. The datawarehouse is the place where you can match all your siloed data . All those small MS Access databases, MS Excel worksheets etc. that you have around generate an inextricable mess to understand what's going on. If you build such an analysis database, you actually have the opportunity to tackle all the different data and reconcile them.

5) Often business people think by categories not implemented in a business application. Your datwarehouse may be the only place where some data may reside thus being applicable to your analysis model.

6) having a different DB, maybe on a different technology, make the analysts feel special compared to all those data entry people, and make it more acceptable to the upper management.

As you see, the point is hardly technical but is related to the way a business looks onto itself. A datawarehouse , to such extent, is like a mirror in front of which you business can realize that there's that little spot on the neck. Do you have examples about datawarehousing in SMBs? Do you share my vision? If yes, can you define it better? If no, why?


# 1

I don't disagree too much, but I feel as though too much focus from consultants goes on building 'the one' database in which everything goes. They can upset individuals within companies who are sometimes 'getting the job done' with those collections of small databases, spreadsheets that pop up around companies big and small. 

Often the focus is about trying to pull data into the corporate database and not enough thought/effort goes into the matching consideration of how to find, access and present the data. Google have demonstrated that its not about storing everything in the Oracle/MS RDB, but about allowing people to find and use it. 

For instance, there may well be a better option for somebody who is currently collecting and presenting data in some 'silo' system; make it an Intranet website or webs-service so that others in the organisation can use it integrated into a back end warehouse.

From  grant  (submitted  10/18/2009 @ 22:53:26)

# 2

Actually, I did not place on visualization and data consuming the emphasys it deserves.

I see the datawarehouse as the first and unavoidable step for having a coherent and seamless data presentation.

More, I repeat, the point is not technical but pertains to management practice. Thinking a datawarehouse is the best way to think about about a model for your business.

From  Stray__Cat  (submitted  10/19/2009 @ 22:36:02)