** This guest post was written by Estelle Nicholson. Learn more about Estelle in the author bio at the end of this post. **
After an initial burst of enthusiasm, we’ve discovered that BI projects, like many other IT projects, get derailed or circumvented for a host of reasons. People who approach BI projects come from environments outside of traditional IT, so it’s important to understand reasons why your project isn’t getting off the ground (or why it’s crashing) and some of the potential consequences.
1. The Goldilocks Project: Making it “Just Right”
Overdesign (better known as Analysis Paralysis). While planning is good, an interminable requirements specification period and inflexible project management methodologies, can cause delays. At the same time, the users’ requirements will change based on what they see in the industry, and the need for the BI solutions will intensify. Overdesigned projects don’t get adopted because, by the time they are completed, the users have moved on to other solutions and aren’t participating in the design and testing process. The lack of end-user involvement means these projects don’t do what the users wanted, don’t have all the data pieces or functionality they wanted, and aren’t flexible enough to be modified. If the developers do get feedback, the redesign results in major components being moved to the “next release” – the death knell for a project.
Underdesign. On the other hand, the allure of speedily getting hands onto the data can lead to overpromising by vendors or IT and disappointment by users. “No up-front design required” solutions promised in the big data world will show you your data. However, to be useful in BI, the design gets pushed to the analytical layer, opening a new set of problems for eager analysts. The skull sweat and design work have to come in somewhere and sometime. Just slapping a BI tool on top of your data just means you’ve paid for a slick new interface and the vendor’s sales team has moved on to the next sucker customer.
Design and implementation time have to be “just right.” Too much takes too long and you lose users’ interest. Too little and your analysts will be floundering in a sea of disorganized data. This requires a delicate balance of user and technical involvement.
2. Lack of IntegrationEven if the design and implementation are right a user has less incentive to switch to a BI solution that doesn’t offer key areas of integration. Examples are when a user still needs to go outside the system to incorporate a needed piece of data, or when a user has a separate sign-on process to use the system. Sometimes seemingly trivial steps in a workflow can lead to resistance.
3. Unrealistic Expectations
Finally, when stakeholders hear buzz terms like big data, sensor data, mobile reporting, etc. it could create a sense of needing to aim higher than what is in range or even than what is necessary at a firm. A firm with nascent BI that is still getting buy-in or figuring out reporting will do better by building up a foundation of BI before worrying about the latest trend.
All of these will lead to people disparaging your BI solution and using something else – leading to…Self-Service BI.
4. Self-Service Circumvention
Masses of raw data obtained through an underdesigned solution or even from an IT department that is churning out extracts to satisfy requests could lead to dangerous circumvention issues. Enterprising, technically adept analysts might maintain a server inside the firewall (at best) or perhaps on the cloud so that they can analyze their data using tools they get from an unauthorized and uncontrolled source.
5. Self-Service Non-Adoption and CircumventionGoing one step further, what about the recidivists who are offered up a BI solution and use the self-service features to simply pull their own extracts of data? This could be a perfectly appropriate use of a BI solution, or it could hog resources and then present the double danger of data in random databases, spreadsheets or on the cloud.
When users bypass approved systems, they also bypass IT’s ability to protect the organization’s networks and sensitive information. In addition to exposing sensitive information, the use of unapproved (and unknown) external services can open a hole for a potential network security breach. When employees leave, you may lose access to those password-protected accounts (even if you knew they existed) – leaving sensitive data sitting out in the cloud for the former employee (and possibly others) to access.
BI can be a game-changing tool for a company if done and used properly, but it’s important to recognize that BI non-adoption or circumvention has implications beyond some wasted effort and money on another failed IT project.
Estelle Nicholson is an independent business intelligence consultant, with experience working on several business intelligence projects at a Fortune 500 financial firm. She has worked on various aspects of (what is now called) BI for nearly 20 years.