TDWI Upside - Where Data Means Business

Five Drawbacks to Self-Service BI

Self-service BI promises data democratization and faster data-based decisions, but it's not without limitations. Here are five considerations to keep in mind about a self-service BI program.

Self-service BI is a powerful tool, but like all tools, it can be misused, especially by people lacking proper training. Most articles focus on the benefits of self-service BI without looking at potential downsides that companies need to watch out for, including:

For Further Reading:

Self-Service BI: Barriers, Benefits, and Best Practices

3 Best Practices for Becoming More Self-Sufficient with Self-Service

5 Rules for Successful Self-Service Analytics

1. A false sense of security

In "Self-Service BI: Barriers, Benefits, and Best Practices," Jim Gallo points out that the original driver of self-service BI was a desire by IT departments to free themselves from effectively doing ad hoc grunt work for different business units. From this viewpoint, self-service BI is not bad per se, but it might lull companies into a false sense of security. Good BI requires solid data engineering and experience in properly interpreting results, something an expert could accomplish with sub-par tools if required but beyond the skills of the average self-service business user.

2. Licensing costs

Like other services falling under the umbrella of shadow IT, self-service BI is not necessarily the most efficient option, especially when departments independently choose different BI vendors. Democratizing analytics capabilities sounds wonderful until the organization realizes that everyone requires a license to do their own BI -- individual departments purchasing different systems hurt the larger organization when it can't take advantage of larger bulk discounts.

3. Ironically, too much accessibility

In "Ad Hoc Analysis: Business Intelligence's Fast Food Problem," we note that the primary attraction of self-service BI (i.e. its ability to make ad hoc reporting both easier and more accessible for employees) can increase the risk of poor-quality reports. Non-data scientists may not go beyond a quick analysis that confirms their suspicions. Worse yet, such reports may subsequently get shared to a broad audience, further spreading misinformation. Trained analysts know to watch out for confirmation bias, the tendency to search for information that agrees with what we already believe to be true. The same rigor cannot be expected from people deeply entrenched in a business issue or those who are under pressure to explain an observed behavior.

4. Painting with broad strokes

Another limitation of self-service BI is the fact that users have tools that can identify general patterns but lack detail. In "What Are the Limits of Self-Serve BI?", Graham Annett points out that "the abilities of the [BI] software are often so broad, that they cannot provide the level of insight that would be helpful or necessary to be applied to increase revenue." In other words, users cannot tease out the juicy details that help companies solve real business problems.

5. All-out data anarchy

Perhaps the most critical of articles against self-service BI comes from Nimrod Avissar, who in "The Self-Service BI Hoax" points out that: "In many cases ... time is not actually saved, because a simple model that would take a professional 15 minutes to build, secure and test takes an end user twice as long, and ends up being passed to IT to add security and make available globally."

Nimrod goes further, pointing out that although self-service BI tools excel at ad hoc analysis, they come up short when used for conventional reporting. Consequently, companies end up dividing their reporting into ad hoc and regular categories and across different platforms. Ultimately this behavior promotes further fragmentation and worsens the issue of data silos.

Recommended Actions

What is a thoughtful company to do?

Realizing that self-service BI isn't a panacea is a great start. Beyond that level of self-awareness, the solution is hardly surprising. Ensure that you have a data strategy and BI strategy. Focus on data engineering to ensure smooth and accurate data pipelines. Employ data experts to extract insights for critical reporting.

In other words, treat your data with the respect that it deserves.

About the Author

Matthew Gierc heads up marketing and business development at 3AG Systems, a data analytics firm that helps companies solve real problems with their data.


TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, & Team memberships available.