The Myth of Self Service BI
January 17, 2012 by Thoughts from the Dataspace
Filed under All, benjamin taub, business intelligence, business intelligence usage, data warehousing
Most business intelligence efforts start with a statement like the following, “Let’s build a BI system so our employees can get whatever data they need, when they need it.” While well intentioned, statements like these are misguided.
The truth:
- If folks do their jobs today without using a shiny, new BI tool, why are they going to put extra effort into learning your snappy new tool?
- Most folks don’t get a big thrill out of accessing data - BI is like a dishwasher, a tool for getting a job done. Do you want to go to classes and learn a panoply of new things just to use your dishwasher?
- Most folks’ data needs are limited to a relatively narrow range of topics (e.g. sales folks typically don’t need to access HR data).
- Even with data warehouses and strong metadata, BI tool vendors have not yet produced a tool that gives anyone simple access to every piece of data in an organization.
The implications:
- A few folks do have the need, desire & technical skills to get and analyze their own data. These folks are called data analysts.
- While it’s not sexy in terms of BI (is anything sexy in terms of BI?), the vast majority of folks have structured jobs that require access to a limited range of data and they usually need it in some set format (e.g. this week’s accounts receivable aging report). Assume that these folks need BI tools that provide predefined reports into which they can enter a few parameters - tools like WebFocus, Crystal Reports, and SQL Server Reporting Services. Also assume that these folks will NOT create these reports on their own - you will need technical experts to create and maintain these reports.
- You are far more likely to see wide usage of your BI systems if your business processes are designed to be data driven. Business process design, along with a culture change towards being data-focused is a prerequisite to widespread BI adoption. It is also hard, time consuming work.
- Widespread BI adoption isn’t always the correct measure of success. A $2 million BI system that serves just one user could be a major success if that one person makes or feeds input into multi-million dollar decisions.
So, if you want a clear path to ROI, rather than starting out to build a BI system, start out to change your organization. Then, figure out how and where BI is necessary to enable that change.
Tying the BI tool to the user
June 29, 2009 by btaub
Filed under All, business intelligence, data warehousing, management reporting
Yes, a 747 and a Cessna can both be used to transport you from point A to point B but, isn’t the 747 a bit of overkill for the pilot who just wants to fly himself to the next airport for a $100 hamburger? Well, in Business Intelligence (BI), many organizations buy a fleet of 747s when all they need is a few Cessnas – they buy tools that are powerful but overkill for most of their users. A great example of this is when a company buys 7,000 licenses of an expensive, powerful OLAP tool, intending to outfit their entire staff with OLAP. Is there a need for advanced online analytical processing (OLAP) in the company? Almost certainly. Are there 7,000 users who are going to slice and dice through their data? Almost certainly not.
You can think about BI needs as a pyramid, small at the top and large at the bottom. At the very top are a few analysts who use data mining tools to identify unexpected relationships and build predictive models by looking at huge data sets (Can you remember when the data mining companies were looking to put mining on every desktop? Mining on every desktop? Really?).
Just below the data miners is another, slightly larger, layer of folks who need to slice and dice through their data – the OLAP users. These folks are looking for things like what products are selling well, in what regions and by which salespeople.
Next is the bulk of the pyramid – the folks in the field who are just trying to get their jobs done. The folks who need BI to execute specific business processes: to see which customers receivables are over 30 days old; to see where maintenance crews have been assigned for the week; to do the actual day-to-day work of the company. Do these folks need to slice and dice through huge quantities of data? No. These folks generally need a set of predefined reports which have a few flexible parameters for users to complete to specify exactly what data to report on.
While the major BI tool vendors sell their tools as allowing users to create their own reports and to slice and dice their data, the bulk of the pyramid never uses this capability. Instead, when these tools are released to users they are released with libraries of pre-configured reports. Most users never do more than use these reports or, occasionally, request new ones.
Once you understand this reality, you start to look at the concept of BI tool standards quite differently. More on this in a future post.
Think you’re overbuying in BI? Drop me a line.
– Ben
