The Paradox of Self-service Business Intelligence

Jul 26, 2017


Self-service Business Intelligence (BI) is a must-have for any institution. It has an important role to play in cutting down on the hours/days spent on the manual culling together of multiple spreadsheets and shadow databases just to provide basic year-over-year enrollment reporting, admissions funnel reporting, graduation rates, course success metrics, retention/persistence rates, etc. If you struggle with these descriptive analyses, there are some software vendors and services providers (ASR Analytics included) that can accelerate your implementation of this essential capability. The data models that power self-service BI can satisfy the requirements for basic descriptive analytics, but what happens when you get a true research question from a new grant or initiative? For example, consider this real-world question:

What is the graduation rate for Degree-seeking community college students in a fall term, excluding current and prior high school dual enrollment students, tracked for six calendar years? A student is 'degree-seeking' if they a.) Enrolled full-time for at least one term within 12 months of starting in the fall cohort OR b.) Enrolled at least half-time for any two terms within 18 months of starting in the fall cohort.

No one can provide an all-inclusive analytics product or a data model that can magically answer the myriad of research questions that regulatory bodies, presidents, provosts, or vice presidents of enrollment or finance might ask you some day. Most of the time the tough questions they ask will cross systems (student registration, finance, financial aid, CRM, etc.) and architectures (on-premise and in the cloud). Unfortunately, at many institutions leadership has been oversold on this self-service capability and IT and IR departments find themselves having to explain why their enrollment cube can’t easily spit out this information. “We spent , why is it taking you weeks to get me this information?!”

So, what is an institution to do?

Invest in analytical skills. Self-service BI is meant to remove your highly analytical staff from doing mundane tasks, e.g. basic enrollment reporting. It is not meant to replace them. On the contrary, you still need them and probably more than ever. The more basic BI shines a light on your institutional metrics the more questions people will have. And as you advance in your analytical competency, many times the questions will not be easily answered by an off-the-shelf data model, as they will be very specific to your institution. No one wants to hear this (and more importantly no software vendor sales person wants to say this), but you probably are going to need to hire more people with SQL and statistical skills (e.g. R and SAS) as your institution becomes more data informed – not less.

Focus on creating flexible data architectures. Cloud applications are great for reducing your reliance on infrastructure, but most solutions will lock up your institution’s data in a vertical application with little hope for being used to answer complex analytical questions.Make sure before you sign with any cloud vendor that you understand how flexible this new system will be when you need access to your data in bulk to incorporate into your data models.

Buy a comprehensive BI platform. All BI tools are not made the same and ones that you get for “free” tend to be very inflexible or are missing important advanced features. While these features may not be needed for simple self-service use, they will quickly become essential for answering the more complicated research questions. Additionally, the pretty/flashy products tend to do only one thing really well. The extra investment you make in a BI platform with a flexible meta data or semantic layer will more than make up for the development effort you will need to invest in a less flexible or comprehensive platform.

Self-service BI and packaged analytical content solutions sold by many vendors are valuable – for about 80% of what you will need. These packages will answer basic questions and remove much of the mundane data combining tasks that are bogging down your skilled analytics users today. They will answer the analytical questions that everyone asks. As your institution matures, however, you will need to start answering your specific and advanced research questions.

Unfortunately, there is no easy button for that.