Building an Enterprise Analytics and BI Practice in Higher Education

This is a tough period for higher education. The industry is faced with financial, political, and pedagogical challenges; however, there are also many opportunities, including the incredible innovation occurring in the data analytics and business intelligence (BI) field. Innovation has long been the heart and soul of institutions, but they are now faced with the idea that innovation should be applied to the institution itself, not solely to the object of its teaching, research, and inquiry. Responding to this situation has been difficult for every institution. As a result, institutions are experiencing increased pressure to better understand themselves in a way that has already been applied aggressively in many commercial organizations and drives the strategy and growth of the most dynamic companies in the world.

How can institutions that have been focused on cost containment in administrative processes shift focus to the investment required to become data savvy, even data driven, given their culture? There are some broad answers and some specific answers to these questions. Let’s tackle the harder questions first.

Culture

Building a data driven, inquisitive, and adaptive culture in higher education is not simple. Until quite recently, institutions were not rewarded for being quick and adaptive. Instead, institutions have been rewarded by reputation and long-standing success built on past achievement of faculty, alumni, and donors. But the world has begun to change. Student bodies are less traditional, state governments are in budget crises, research is increasingly privately funded, students are often more tech savvy, and the world is online and is certainly a smaller place. This has led to new modes of instruction, more collaborative research, and increasing competition for both.

All of these changes have directly confronted the antiquated, internally parochial, and slow pace of change common in higher education. Universities and collaboratives in which they participate have begun to find ways to align and analyze data for action. As the data emerges, it will be up to institutions themselves to adapt to what they find.

How can institutional leadership prepare to pivot without limiting academic freedom or shared governance? This requires quality data and analysis, trust, and decisiveness. Is there a roadmap for this journey?

There are some basic activities and methods that can bring institutions forward. Outlined below is a recipe that has worked in several institutions but, as any good recipe does, varies from kitchen to kitchen.

It Starts with a Question

What burning question at your institution brings varied answers to the decision table? This a good place to start, by focusing on one important question and expanding the scope from there. An example might be one of the metrics that drive budget allocation. Can leaders agree on one set of metrics and source of data to drive that process? Can these leaders gather a team to build the infrastructure together in a sustainable way? This is a place to start.

A Partnership

Partnerships can begin with a small group of leaders that cross academic leadership, analytic, and technical areas of the institution. It is important for these leaders to have a keen focus on outcomes rather than ownership. When institutional leaders partner on the critical components of success and start small, there is a real change to build success. Institutional research, academic leaders, and IT must bring their skills to the table, agree on an investment, and plan to produce the most critical analysis to drive institutional decisions.

It is important to note that, for many, the start of this process does not require deep data engineering and data science skills. Basic data management and analysis skills can provide results that drive action while building the cultural trust of a wider program. Skills can be augmented over time as scope expands and the complexity of questions increases.

The Technology

Though building an enterprise analytics practice is not just about the technology, a solid technology infrastructure is required. These technologies serve the question and are not the answers themselves. This does not imply that a CIO cannot lead these efforts; CIOs have many talents, and often data and analytics are core to those skills.

The current technology climate is not easy to evaluate. If an institution does not already have major investment in analytics and BI, it might be best to look at where technology vendors are investing most heavily. As in ERP five years ago, there is little visible investment today in on-premises analytics technology. So, a path to the cloud must be in the medium-term plans for technology organizations, or they will be faced with many of the questions that application support teams face today. The cloud is proven to be a great place to experiment and innovate, and it is quickly becoming mature enough for enterprise class analytics platforms.

With that being said, the on-premises data warehousing and analytics technology that is used across higher education today is likely sufficient to provide answers to an institution’s first set of questions. Institutions with legacy technology and as little as two to three focused staff members have been able to start to change culture and exercise the data-driven decision muscles needed to scale a successful program. This success can drive planning and investment on a broader program.

The Recipe

The road ahead of that initial start is multifaceted. While required skillsets may already exist within some institutions, many will have to either train staff or recruit new staff. Many institutions point out that the required skillsets may already exist in the research community on campus but are not aligned to this effort. Borrowing from or learning from that set of disciplines can be valuable.

Skillsets

Data governance and data quality skills are focused on ensuring that the data used to answer key questions is understood, managed, and transparent, with the goal of ensuring trust in the data.

Data engineering and data science ensure that data is moved, combined, and technically analyzed to achieve the correct outcomes without bias. These skills are focused on trust in the analysis itself, avoiding misinterpretation and poorly formulated results.

Front end/user experience design, a skillset often overlooked in analytics efforts, focuses on ensuring that the data and analysis are communicated in an effective way, conveying the message intended and removing bias from the formatting of the outcomes.

Platform/backend engineering skills are required to ensure that the technology is managed efficiently, with high assurance, repeatability, and extensibility.

Finally, project and program management skills are required to ensure the smooth functioning of the work, especially across skillsets, and that the program remains in alignment with strategic goals, and to communicate effectively with stakeholders.

These skills are required for a fully functioning program to proceed, but again, not every skill is required to start. Individuals with multiple skills can be used to help get a small team started.

Technology

Selected technology needs to be reliable and your team must understand how to leverage it. However, as an institution’s plans evolve, there are some specific technology capabilities that need to be scoped.

Stable and scalable technology is required to move an enterprise BI/analytics program forward. Since building out a data warehousing infrastructure with high-scale compute and storage requirements can be extremely expensive to purchase and deploy on premises, many institutions are starting to focus on building out these capabilities in the cloud. The leading cloud providers have similar capabilities; choices often come down to compatibility with other technologies within an institution and/or cost factors.

As the process moves forward, data lake technology combined with more traditional data warehousing technologies provide great flexibility for faster data discovery and analysis. The results can lead to long-term, high-accuracy, and ongoing analyses typical of data warehousing that are often required for long-term management of a function. Enterprise data does not have to be reformulated into a data warehouse schema to perform analysis. Regardless of whether big data sources are being utilized, a data lake can simplify the intake and quick analysis needed to be innovative and iterative. It does, however, need to be managed, or it can become unwieldy very quickly.

Data visualization tools make quick work of efforts that were very difficult with older tools. These tools help visualize large amounts of complex data to get to insights faster. These capabilities are moving from specific visualization tools into other more general tools—even ERP embedded tools. Still, there is a strong need for operational and managerial reporting that is grid-based. Tools for easy exploration and manipulation of data is critical.

Lastly, all of these tools require an ability to share data and analyses without being completely unmanaged. Emailing spreadsheets and PDFs is not good practice, nor is it efficient. It also does not align well with good data governance, where everyone understands the derivation and quality of the data they are consuming.

Leadership

Finally, the hardest part, right back to where we started: What does leadership need to do?

Define your critical questions. Prioritize the critical questions and understand that one criterion included in that prioritization is the cost to get the answer. For example, if #8 on your list can be achieved first, is it not the highest ROI? With a focus on building trust in the data and the process, quickly making key decisions from the analysis is a key criterion for priority.

Shared governance must be aligned with the critical stakeholders, including those functional areas that will be impacted by the decisions made. The defined governance should agree on the measurable success criteria of the effort as it begins—as well as the intended resulting decisions that are aligned to it. Without outcomes that drive action, analytics are just “interesting.”

Lastly, leadership needs to align resources to the effort, with enough time, staff, and capital to do the job well. The team should build components that can be reused in later analyses and not perform on-off analyses. Over time, the data components that are part of the data lake and data warehouse can grow, including more and more cross-functional data elements that can be used for wider, more complex analysis.

Conclusions

As in any modern technology effort, business intelligence and analytics should not be assigned only to IT, nor to any individual unit to own and run with. They are by their nature collaborative and need to be managed as complex efforts that affect and are affected by the various parts of the institution. Though this is often difficult territory for institutions, the collaboration needed to move forward will also drive the innovation and insights achieved. Higher education, as an industry, is unique and uniquely challenged at this point in time. The industry needs to find ways to better understand and express our value, opportunities for improvement, and impact on our communities. Higher education needs to innovate its own methods, in how it educates, performs research, and connects to our communities. These efforts will require the same discipline and shared governance that have built our institutions, but we must adopt new patterns and processes much more quickly than we are used to doing.

The views and opinions expressed in this article are those of the author and do not necessarily reflect the official policy or position of The Tambellini Group. To express your views in this forum, please contact Hilary Billingslea, Director, Marketing Communications & Operations, The Tambellini Group.