Today’s Tech Landscape Calls for Higher Education to Revise Its Data Strategy Approach

Over the last several months, I’ve been struck by the conversations I’ve had with college and university CIOs and CTOs about data.? Regardless of the size or type of institution, or their levels of structural complexity, these technology executives expressed an urgency to integrate data from across the institution’s many information systems.? These conversations marked a shift in tone from the last several years, where most IT divisions were largely satisfied with the strides made to bring data together into physical or, later, virtual warehouses.? Those earlier conversations were almost the opposite in scale regarding urgency to act.? Other priorities were well ahead of data integration, and existing integrations were a source of pride.? What happened to create this urgency?

The Goalposts Have Moved

One of the major reasons for the tonal shift is that the goalposts have moved downfield.? Or, removing the sports reference, expectations for the use of data have significantly increased.? Until recently, much of the data strategy for higher education institutions was focused on reporting and compliance.? Whether to forecast finances, enrollment reporting, compliance with state or federal data requirements, or to better understand the scope of the student experience, getting access to a complete set of data, and access to a stable set of data for reporting (production data is ever-changing) was a significant challenge.

Significant efforts were devoted to the plumbing system required to populate a growing suite of point solutions, often purchased by departments, that were designed to accomplish a narrow set of tasks.? Moving data from the system of record (SIS) to these many systems required timely data transfers.? Once implemented and in production, those users were concerned that their data “didn’t match” the reports coming from the SIS, and, well, you get the point here, back to the focus on reporting. Data strategy emerged to solve these problems.? How can we create a repository of information that provides a reference for these seemingly disparate data?? What tools are available to help end users understand the data that exist?? How can we train end users on the definitions of data and how variances in those, as well as business process gaps, account for many of the perceived data errors?? As a result of these strategies, institutions were able to create more solid portraits of institutional data, largely accounts of what happened in the past and what could be reconciled between systems.? Snapshots, data cubes, and lots of SQL report writing was required to establish and maintain an increasingly complex plumbing system of data.

The new goalposts are focused on real-time (or close to real-time) data, placed into the hands of end users and into tools like CRM and data visualizations, fueled by Artificial Intelligence (AI).? Our past focus has been on lagging indicators, those data used to analyze what happened (admissions yield, retention, completion, etc.) and to develop supports, communications, experiences, to improve outcomes.? These initiatives have been general in nature, offering services to groups or to everyone, in hopes of intersecting with those who may need them.? The data focus is moving to leading indicators, those that tell us what is happening right now, and to whom, so that we can respond with personalized, laser-like accuracy.? Advances in computing speed and power allow these innovations to be realized, but only with new data strategies.

Needed Data is Trapped

About 25 years ago, there was a strong movement in IT to purchase an enterprise resource platform (ERP), which promised to provide a comprehensive picture of needed data across the student, faculty, employee lifecycle.? Sprinkle in a little Y2K anxiety, and these ERPs became the norm for higher education.? The skill sets and infrastructure to support ERPs was grown from those perceived needs.

What didn’t occur was a truly comprehensive data system.? Varying by ERP, most lacked housing, library, co-curricular, career, payment, parking, and other systems required to operate the institution.? What couldn’t have been foreseen at that time was the rise of online programs, LMS, CRM and many other systems that became essential to a modern college environment.? While the core of the ERP – enrollment, employment, finances – is rock-solid, the additional systems continue to increase, to the point where it is common to find 35, 50 or 100 additional systems in play today.

In most instances, the ERP populates the initial data, so that these ancillary systems contain the accurate student bio-demographic data, although manual data entry is possible in many, creating potential identity clashes and duplicate records.? These additional systems then grow their own data, which exists only in that system, and remains “trapped” there.? Much of that data represents information on how students are doing at that moment – interactions, successes and failures, experiences, and community.? While some may refer to these as “data silos”, it may be better to call them “cylinders of excellence.”? Each is assumed to be maintained for a specific purpose, and with care for the information growing inside them.? Faculty and staff consider them vital to their success, as they contain information and insights they could not otherwise obtain.

This “best of breed” approach (everyone gets the system that works best for their somewhat unique needs) has come with significant downsides.? Aside from the irritation and inconvenience of logging into multiple systems to accomplish required tasks, the student, faculty, and/or staff experience is sliced into many pieces.? It is difficult if not impossible to create a complete picture of any given person.? Warehouses took us some of the way toward a unified picture. However, the extract/import routines required to establish and support a warehouse were Herculean in scope, and much of the data on student success is not included.? Leading indicators, harmonized in real-time and available to end users with ease, seems like science fiction.? Application Processing Interface (API) technology, data lakes, advanced middleware tools, and more have brought us into a new era of opportunities, and new investments. Strategic approaches are essential to a reasonable and effective level of resource investment.? How can these trapped data be liberated and turned into insights and action?

System Streamlining

One step toward better data access and utilization is to review the myriad systems deployed at a college or university.? Is it? possible to provide the needed functionality while reducing the number of systems that are producing and trapping data? While the upsides of this are quite evident to IT offices, they are less appealing to end users of those ancillary systems.

With a strategic goal of having more data available for use across the student lifecycle, fewer systems means fewer integrations to set up and maintain.? Some of these systems require maintenance and routines to import data from the SIS for population of records, and some also require resources to set up and maintain exports to a data warehouse.? Unless IT resources are unlimited (a rare scenario), this means that there are only so many of these integrations that can be reasonably managed.

The significant issue in this step of data strategy is cultural change.? Trust is at the center of it, where end users (departments and programs) know what their systems do for them today, and may be leery of a new system and promises that functionality will be covered.? Often, they seek peer references to gain confidence that this move is worthy of consideration. Institutional memory can be long, and any past negative experiences with moves from one system to another are likely to be used as rationale to resist system consolidation.? It may be necessary to identify an entrepreneurial department, school or program that is willing to pilot a new system, then evangelize its merits to other faculty and/or staff.? Tasking departments to leverage their peer networks or professional meetings to speak with colleagues who use the proposed technology may ease anxieties.? Often, the company providing the technology solution may have references, and when executed well, these interactions can foster some “real talk” about how it has worked and the benefits and challenges of moving to a new system.

Financial incentives may be possible in system consolidation.? If the department that “owns'' the ancillary system will no longer need it, there may be some budget to be used for other department needs.? This may or may not come to fruition, as new costs of licensing the new system have to come from some pocket of the institution, and there are several institutions that assess a “tax” on the department for hosting and managing this new software.? These arrangements often incur greater scrutiny of the software’s performance, especially in comparison to what was used in the past, and they may slow or limit adoption of new technologies.

Overall, this is a logical step and, even with its challenges, system consolidation should be strongly considered in new data strategies.? Maintaining dozens of independent systems is costly, and limits the power of new tools when their data cannot be made available to them.? Those institutions that already have a more centralized approach to software procurement and maintenance will have a significant leg up on their peers.? Institutions that have succeeded in this work have identified the systems that can be replaced, then worked over time with departments to sunset existing ancillary systems at or before contract renewals, prioritizing those that most need it, and those departments most ready to make the transition.

Say Hello (or Welcome Back) to Data Governance

The administrative owners of the SIS, usually spearheaded by the Registrar, are used to data governance work, even if it is not formally called that at the institution.? Routines to identify duplicate records, work with the admissions office to ensure that needed data is input into the system in common formats, limitations on access to edit tables, etc., are all in play today.? These have been necessary for the maintenance of student records, the specialty of the Registrar but shared work among admissions, financial aid, student accounts, HR, and anyone else whose data becomes part of the SIS.

Around 2014, there was significant interest in bringing additional data into student records.? These data were from learning experiences (internships, global studies, research, leadership experiences, etc.) that were governed by faculty but executed outside the academic course structure.? Even if there was representation of the experience on a transcript, there was no information that told anyone else what that experience entailed, and how it prepared the student for careers and studies beyond the baccalaureate.? Descriptive data could supplement, not replace, academic course data.? As the work progressed, some innovative institutions worked to bring the data into a common site, so that it could be used alongside the SIS academic program data.? They discovered that extending data governance routines to academic and co-curricular offices was necessary to ensure the same quality standards expected from academic learning experiences (courses or modules).? The scope of data governance extended to career services, global studies, student life, and other departments.? Overall, these were incredibly successful, and the offices became more aware of the need for data governance, as a result.? They saw how their data impacted other areas of the student experience, and were able to adapt their routines to ensure quality information in these new student record types (Comprehensive Learner Record, Learning and Employment Record).

This example holds clues to how data governance will need to evolve and expand, as part of a new data strategy.? The scope of governance must go beyond its traditional bounds.? In some instances, strong data governance routines have only been in place when a new major system has been implemented.? All too often, the groups that met and worked out data governance requirements for that system disbanded after implementation.? Rather than ad hoc status, data governance must be a standing workgroup or committee of the institution.? It will need to raise awareness of data governance needs and educate new departments, offices and users on their roles in this expanded governance structure, so that they can fully and successfully participate in it.? Real-time data, required for leading indicators of student success, create even greater needs for this than those data that come from completed experiences, or past achievements.

There are some excellent resources on data governance available from EDUCAUSE[1].? Gartner also has a good, fairly short article on this topic[2]. These should be consulted when considering a formal and/or informal approach to data governance at your college or university.

Setting Sights on A New Data Paradigm

The ideal data environment is one where the data flow from the very first time a student has an interaction with the college or university, to decades beyond their completion of a credential.? This stream of data is capable of pulling on the entire student history and, as it grows, becoming more nuanced in how it can fuel a personalized experience for the individual, and how it can inform our understanding of the entire student/alum lifecycle.? As a side effect, this larger, comprehensive data set can foster rich AI analyses and insights.

We can start the paradigm in the recruitment phase of the life cycle.? Here, the institution seeks to attract various student personas (direct from secondary, transfer, adult learner, graduate, international, military, etc.), then provide personalized information to them.? Over time and through increasingly personalized information, the institution hopes the student will learn that there is (or isn’t) a clear and compelling case to pursue studies.

The model for data collection and use can be formed around three key questions:

1. What interests me?
2. What am I doing/what are my actions?
3. To whom am I speaking?

For each question, there are data sources that provide insights beyond the typical CRM data.? Figure 1 illustrates those and joins them to the CRM, where they can be analyzed and used to increase personalization of the admissions journey.

Figure 1. Map of Data Sources for Recruitment and Admissions

Students are not always clear about what they may want to study.? There may be several areas of interest, and decisions may be made about the combination of academic and co-curricular offerings, or the availability of certain program schedules.? Traditional inquiry forms and cards may not adequately capture these combinations, and interests may evolve or shift over time.? Time spent on a particular web page (i.e., financial aid and payment plans, evening and weekend course schedules) may indicate the intensity of interest in that topic.

Before we dive into the strategic questions to ask about data and source alignment, we’ll consider the current student data environment, which is much more complex than the one for recruitment.? Figure 2 shows a limited number of systems that may share some bio-demographic, term, or other data about a student.? Each also generates much more data that could indicate the immediate condition of that student’s success.

Figure 2. Map of Limited Data Sources for Current Student Enrollment

The red arrows indicate the typical position of data transfer across systems.? The Student Information System, considered the system of record, pushes the accurate student data into other systems, so that they can have that data available to use in student interactions.? Here, the data remains “trapped” and unavailable for use in analysis, save for tedious, manual operations to extract, join, and analyze it.

The aqua lines represent the liberation of those data from the systems.? Because these data have grown significantly, and may exist in both structured and unstructured formats, their first stop is a tool that can integrate and harmonize the data.? According to Gartner[3], “Data integration tools enable organizations to access, integrate, transform, process and move data that spans various endpoints and across any infrastructure to support their data integration use cases.”

Imagine, though, that these data have come into the tool in various formats.? Because many “point solutions” (applications or software that provide a specific solution or perform a specific business function) allow users to be manually added, there may also be differences between the data from the system of record and the point solution entries (names are one of the biggest issues here).? Data harmonization tools work to reconcile differences in formats, units, and structures, and create a unified data set (single schema). Reconciliation of identity can also be performed at this stage.? This allows the data to be more accurately used by the downstream tools in Figure 2.

The figure shows the data flowing into a lake.? This could be a warehouse, assuming the data have been successfully harmonized at the integration stage before it.? Lakes can store data in structured and unstructured formats, where a warehouse usually requires that the data be uniform and ordered, prior to or at the stage or loading.? This is a simplified explanation of what can be a complex issue.? There is a fairly straightforward Coursera explanation[4] with tables that outline some of the key differences between data warehouses and lakes.

The Alumni Journey

Colleges and universities want their alumni to stay engaged.? Traditionally, this has been to harness their time, talent and treasure to benefit current and future students, and to support initiatives of the institution, financially and otherwise.? Here, the greater the understanding of the student experience alumni offices had, the greater the opportunity to personalize their engagement, post-enrollment.? Today, we see the role of lifelong learning as an additional, important element of that engagement.? The ability to offer reskilling or upskilling to alumni holds value to them, and to the institution.

From the first moment they knocked on your website to see what you offered, through the scholarship, academic and co-curricular experiences they had while a student, the data that inform personalization now become available.? Layer onto this the engagement data of all alumni events, campaigns, communications, and the picture of the alum post-graduation becomes clearer.? Their affinity to programs and organizations, their experiences while a student, can inform their potential choices for alumni engagement, continuing education, and philanthropy.? Fueled by AI, the Advancement organization can now see patterns that help it better and more effectively communicate and cultivate alumni engagement and donations.

What data matter here? We can start with how students are affiliated with their major, the groups they joined, the scholarships they received. It is such a new frontier, however, that the correct approach may be to utilize neural networking in AI to discover the associations that exist but that we may not yet see.? In short, pour it in, stir it around, and see what comes out!

Conclusion

In an ever-increasing world of data creation, understanding what data matter, and how they can serve higher education, is incredibly important.? New tools, such as AI, have emerged and rely upon data that makes them more intelligent and accurate.? Reliance upon and expectations for data visualizations and reporting are high, and the patience for joining spreadsheets or waiting for someone with SQL skills to write a report is thin.? Seeing only one part of the student experience is limited in its approach to understanding and personalizing the journey for students.? How do these new realities inform the strategic goals that may be part of a new data strategy?? Here are a few potential strategic goals to consider:

1. Identify and join the data that are most relevant to the student journey, based upon the student’s position in lifelong learning (thinking about attending, already a student, alum)
2. Consider system streamlining, and winnow down the number of data sources while maintaining a great user experience for students, faculty and staff.
3. Identify and acquire the right tools to harmonize the data across the many systems that hold pieces of that student journey.
4. Power insights using advanced tools, such as predictive AI, to understand the complexity of the student experience, when viewed comprehensively.
5. Utilize tools, such as generative AI, to personalize the experience of prospective, current and former students, meeting their growing expectations for information and experiences that are “just enough, just for me, just in time.”
6. Place data tools into the hands of end users – advisors, recruiters, front-line staff, teaching faculty, administrators – that provide them insights on how to be more effective in addressing the needs and preferences of students they serve.
7. Create and implement training programs for end users that provide ongoing assistance to them, as they use new and vastly improved, more complex data tools.
8. Implement an institution-wide data governance team or committee, charged with oversight of policy formation, training and issue resolution, with the explicit outcome of greater understanding of and more effective data governance.

If there is a single word to use when forming new data strategies for higher education, it is “activation.”? Too much of our data is static, trapped or siloed.? The goal of new data strategies must be to activate that data in ways that place them into the hands of end users, not researchers or administrators, and into tools that help them personalize and serve your next, current, and former students.

References:

1. https://er.educause.edu/articles/2024/6/you-cant-have-digital-transformation-without-data-governance
2. https://www.gartner.com/doc/reprints?id=1-2HW2D5T2&ct=240620&st=sb
3. https://www.gartner.com/reviews/market/data-integration-tools
4. https://www.coursera.org/articles/data-lake-vs-data-warehouse