Friday, November 18, 2011

The Biggest BI Blunder

Many Business Intelligence projects struggle and fail. But there may be just one general reason why they go south: the individuals in charge did not treat the initiatives as software application development projects.

Instead, many people with failed projects saw BI as a business initiative or as a financial exercise and ignored decades of best practices of software development lifecycle.


If you try to run a BI project as something other than a software application development effort, then you are immediately starting off on a bad trip with the wrong people on the wrong vehicle.

Wrong Approach
By not understanding that your BI initiative is an SDLC project, you will not follow well-established practices that would improve your chance of success.

You might not even follow basic project management concepts such as the "Magic Triangle" to define the "scope" of what you promise to deliver by what time and at what cost. Experienced SDLC developers understand that in a project you are stuck with three interrelated variables.

Changing one of the following impacts the others. 
  • Deadline (the length of time to complete the project)
  • Functionality (the scope of the project, the features your provide)
  • Cost (the resources on the project)

If you need your BI project to be done quickly, then you must add resources (increase cost) and/or reduce what will be delivered (reduce functionality).

If you want lots of functionality, then you must push back the deadline and/or add resources (increase your costs).

If you need to lower costs, then you have to be willing to eliminate functionality and/or push back the deadline and work with fewer resources. 

Ka-Boom!
Experienced SDLC developers also know the difference between the "Big Bang" approach and a phased, iterative roll-out.

BI initiatives are sure to fail when an influential person gives vague marching orders such as, "I want an enterprise dashboard where everybody in the company can access all data in any way they want. And I need to demo it at the annual conference in three months. Oh, and make sure it works on my iPad, Bob's Android phone, and Sally's iPhone, and Greg's BlackBerry."

Trying to create an entire complex application in one fell swoop is a sure recipe for failure. 

Burning Your Dollars
Successful SDLC projects follow a phased staffing approach. Not all individuals needed on the project should start on Day One. Yet that is a common mistake on failed BI projects.

Imagine building a house. On Day One, you only have a plan and an empty lot. The first step is to dig a hole and pour the foundation. If your custom home builder told the electricians, plumbers, dry-wallers, and painters to come on Day One and just have a seat in lawn chairs to watch and wait for their services to be needed, you would immediately recognize this as a waste of time and money.

Yet bringing in GUI dashboard developers to sit and wait for the database team to finish a repository is common on failed BI projects. It just is not as obviously a waste since IT people look more productive sitting in a Herman Miller chair in a cubicle rather than out front on a lawn chair. 

Wrong Expectations
Experienced software developers know they will not be successful unless they first have a clear definition of the application they are to build. Successful individuals are able to take a fuzzy idea from the business and work jointly with others to properly define and formally document the specifics of what will be done.

The worst BI projects are those where the ideas for the dashboard are in one person's head. This person is always frustrated that nobody else on the project is smart enough to understand what she is thinking. Without formally dumping ideas from that person's head into a well-articulated requirements document, the BI project is doomed. Yet it often happens.

Note: read the "Tappers and Listeners" section in the Heath Brothers' book "Made to Stick" for more on this topic.

The first problem of not knowing what you are building is that you will create the wrong thing; your project has to fail since it can never meet expectations.

There is another problem of not clearly defining what you are building: you will never be done. At some point, the executive paying for the BI project will start to scream about the time and cost. "Why is this BI project taking so long?"

Well, it was never designed with an end goal in mind and as a result the BI developers are continuously running a marathon on a circular track.

Wrong People
If you do not understand that you are running a SDLC project, then you will not employ the right people.

If you view this as a business initiative, you will naturally pick people who understand your business. It just seems to make sense to have the guy in Finance who calculates your business metrics to run the BI project. But without a formal SDLC understanding, Bob in Finance is not going to be able to run a successful software project.  

Repeatedly, I see a certain type of organization fail in BI initiatives. These have been those where their IT groups have traditionally done operational support; they do not perform application development projects. While they are great at running the networks and keeping e-mail and packaged applications running, they are not software developers. 

When these types of IT groups get complex BI projects thrust upon them, they typically fail. This is not what they know how to do. 

They should not be blamed; their organization is intentionally designed to provide operational support.

They are inexperienced in software development. Yet it does not occur to the executive that he or she is asking for the impossible from these employees. Instead, the project fails and the executive asks, "Why can't my IT development implement a dashboard?"

Plan for BI Success
Business Intelligence is a software application development endeavor; without that basic worldview in place, your BI project will fail. Instead, you will jump into a BI project with the wrong approach, the wrong expectations, and the wrong people. 

A "Just Do It" command from an executive will not make an enterprise dashboard magically appear. To be successful, you will need experienced people following the right methodologies and best-practices to create your clearly-defined BI application. 

Thursday, November 17, 2011

Boris Evelson Predicts BI Future

Well-known Business Intelligence industry analyst Boris Evelson documented his Top 10 predictions for the 2012 BI market. You can read them on his blog.

Boris foresees a continuing trend of end users being able to serve themselves with better BI applications. To meet these needs, software vendors and IT groups must continue to innovate and provide what is needed for self-service BI.

Boris mentions innovations that you would expect, such as mobile BI, big data, BI-specific databases, and the cloud.

I found "Exploration" on his list to be interesting. That must mean "whatever you want to do" type of interaction with the data, going well beyond slicing and dicing a predefined cube structure or painting an ad-hoc report from a simplified data view. Boris seems to be pointing to "limitless BI" data discovery.

I also like Boris' comment about how enterprise standardization on a particular BI product or technology stack is not taking place. Instead, Boris says:
"Enterprises will learn to live with multiple BI tools. Forrester client inquiries about how to live with multiple BI tools far exceed inquiries about platform consolidations."  

As always, Boris makes good points.

About Me

My Photo
Helping companies make better decisions via Business Intelligence. INTP working on the E&J. Traveler, reader, family guy, coffee drinker.