Is a data warehouse required to implement business intelligence?

Hi,
I always recommend having a good data warehouse before implementing BI. Reasons are that your company's data are probably in somewhat messy condition and/or your company have multiple data sources. If you want to ensure "single point of truth", the best way is to get all those data from databases, spreadsheets etc into one single well-organized data warehouse and in the procesou can even "cleanse" the data.
Then all reports or analyses rely on single source and you can't have the situation in the meetings that everyone has "his truth", his numbers.
There are BI tools that don't need classical data warehousing (like Qlikview), but nevertheless I would always reccomend one.

You don't NEED a data warehouse, but as Hrvoje says, it can be in your interest to have one. There are plenty of BI solutions out there (notably, bime) that allow you to connect to and mix several different data sources at once.

I agree with Odusina and Hrvoje that a data warehouse provides a much stronger basis for analyzing the data, especially when dealing with multiple data sources.

That said, a data warehouse is generally a fundamental part of a BI solution. If you buy a traditional software BI solution, a data warehouse is one of the components that you need to integrate. I work for a SaaS BI company, and we create data warehouses on the fly for you as part of the offering.

So in some cases, buying BI means that you've already bought a data warehouse. (There are some purists out there who would argue that if you have a solution without a data warehouse, you haven't really bought BI. You have a data visualization or reporting solution only.)

- Barbara
Birst
www.birst.com

You MUST have a quality data source for data-driven business intelligence. The data source may be an operational data store, a flat file, a columnar database, or a data warehouse structured as a multidimensional or relational database.

The key to business intelligence is having some data that is worth examining and a tool to examine the data and display results. I suggest you ask the following:

1) What do we want to know?

2) What data do we have that may provide some answers?

3) hHow is the data currently stored?

4) Do I have or can I purchase a tool to analyze the data in its current format?

5) Is this a cost effective solution?

6) What alternatives are available?

This is the big question that is always put forth by the business side to IT.

Some points on Data Warehouse Initiatives.

- They can often be rather lengthy when approached from the whole enterprise perspective. This is often how many DW architects want to start.

- They CAN be done in small segments of business need IF the architect does a good design planning for the whole enterprise DW foundation. But this is often overlooked, not allowed, or not pursued.

- A lot of DW initiatives can be rather costly.

- More often than not, time to market and return to the business is slow to appear during the DW effort.

- Because of the above, Data Warehouse initiatives have a propensity for failing.

Now for your question... Do you NEED Data Warehouse? No, you don't. As mentioned in some of the previous posts you do need good structured data. This does NOT mean you absolutely need a DW. IT tends to forget that the business side has been doing ad-hoc BI for years using Excel and Access. It has just never been called BI :).

What I have seen working is the following approach.

-Get the tools that enable greater flexibility and capabilities for BI into the hands of the business users. (If they use Excel/Access and pivot tables/reports then they will love something like PowerPivot)

-Let them save their ad-hoc reports using the tool and provide it to other units/divisions for consumption. (They are doing it for their needs already, make the effort available to the enterprise for greater communication and accuracy)

-IT should monitor the usage. See what trends there are and the usage that is taking off with the business. Then IT should target those datasets to provide enhanced data with DW.

The above would let a business grow into a DW by first letting them get excited with what tools will let them do and then they would be more willing to commit to DW initiatives to enhance their usage even more.

There is an upside to both ways... Typically data stored in your ERP system is not ready for BI reporting as it exists in the ERP and needs modeling. This can be done but not without a cost. The performance hit on ERP processes and the performance from a BI perspective on reading through potentially millions of records can be costly with regards to IT dollars. On the flip side the Data Warehouse methodolgy is typically not "real-time" which can be highly desirable in a Business Intelligience environment. The upside to the Data Warehouse is that it allows for controlled data access and the modeling process and transformation can be done through the ETL and the Data Warehouse