Mark Maby

Senior Research Director, Industry Practice
Info-Tech Research Group

Your Challenge

Higher education CDOs face a long flat list of data use cases and no defensible basis for choosing between them.

• Operational areas compete for the same limited resources: student success, research, finance, facilities, and workforce.
• Planning cycles run 12-18 months; the landscape shifts before priorities are acted upon.
• Every use case looks equally compelling without a strategic anchor.

Common Obstacles

Feasibility, strategic value, and cost are each assessed informally and never combined.

• No shared evaluation framework: initiative lists substitute for strategy.
• Data readiness is overestimated: fragmented platforms and inconsistent metric definitions mean use cases that appear executable stall in production.
• Unweighted portfolios dilute resources no institution can afford a portfolio shaped by inertia.

Info-Tech’s Approach

A structured method for producing a defensible prioritized shortlist in a single working session.

• Anchor to strategy: six weighted higher education value drivers.
• Create a common currency: three dimensions combined into a single priority score.
• Deliver a shortlist: ranked output and quadrant chart ready for a data strategy.

PLANNING CYCLE MISMATCH

12-18 mo.

Traditional planning cycles rarely survive contact with the pace of technological disruption.

Source: CFAR

FINANCIAL PRESSURE

64%

of UK institutions project deficits by 2026–27; a potential £3.4B net income reduction sector-wide.

Source: Consultancy.uk

NO SHARED EVALUATION FRAMEWORK

“A plan is not a strategy.”

Roadmaps become exhaustive lists that offer the appearance of direction but lack an integrated core of choices.

Source: CFAR

ANALYTICS TALENT GAP

78%

rise in demand for higher-ed data scientists since 2015 — against institutional researcher salaries of $60-68K, well under the $133K market rate.

Source: Harvard, 2020

1: Select Data Use Cases From the Library

Select use cases from the library to build your longlist

A longlist anchored in real use cases, not a wish list

2: Establish Your Strategic Value Drivers and Scores

Configure value drivers and score each use case against your institutional priorities

A strategic value score that reflects your institution's specific priorities, not generic best practice

3: Assess, Prioritize, and Act

Rate feasibility, estimate costs, and interpret the ranked output

This step involves the following participants:

• CDO/CDAO
• Select data team members
• Line-of-business stakeholders (recommended, to confirm real institutional relevance before entries are locked in)

Outcomes of this step

A longlist anchored in real use cases, not a wish list.

INSTITUTIONAL PROBLEM

Advisers cannot manually monitor every student's behavioral signals across systems. At-risk patterns — missed classes, declining grades, reduced LMS engagement, deteriorating aid balances — surface too late for a timely intervention. The result is preventable attrition that costs the institution retention revenue and costs students their degree progress.

HOW IT WORKS

A predictive model built by mining 2.5 million historical grade records scores roughly 40,000 enrolled students every night against ~800 risk indicators — course performance, major-fit signals, fee balances, LMS engagement changes. When a student crosses a risk threshold the system pushes an alert to their assigned adviser, who has a window (historically 48 hours) to reach out. The system does not intervene directly; it routes scarce adviser attention to the students statistically most likely to need it, before they self-report trouble.

DATA PREREQUISITES

• Student enrollment and demographic records (Domain 1)
• Historical grade and course completion data (Domain 2 / Domain 3)
• LMS engagement and attendance data (Domain 3)
• Financial aid balance and disbursement data (Domain 8)

PRIMARY ROLE OWNER

Provost / VP Student Success

EVIDENCE

Georgia State University deployed GPS Advising with EAB, monitoring 800 risk factors daily across 40,000 students. Since 2012 the program has driven 250,000+ alert-prompted advising meetings. Six-year graduation rate rose 23 points since 2003; four-year rate up 7 points since 2012. Achievement gaps by race, ethnicity, and income were eliminated; bachelor's degrees to African-American students up 103%. Tuition savings reported as $12M/year (NPR, 2016) or $21M/year (GSU).

REFERENCES

• “Approaching Student Success with Predictive Analytics,” Georgia State University
• NPR, 2016

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Students persist in majors they are unlikely to complete, consuming semesters and financial aid. Poor major fit typically surfaces in year three or four — too late for a cost-effective intervention. The institution absorbs the advising cost of a student who will not graduate; the student incurs debt without a credential.

HOW IT WORKS

Every major is mapped to a sequence of critical requirement courses and milestones that must be completed on time to remain degree-eligible. Each night the system recomputes every student's on-track or off-track status against that map. A student who fails a critical course is flagged as unlikely to succeed in that major and directed to meet an adviser before further registration — surfacing poor major fit in year one instead of year four.

DATA PREREQUISITES

• Student enrollment and academic standing records (Domain 1)
• Course completion and grade history (Domain 1)
• Major critical requirement definitions and course sequences (Domain 2 – Academic Programs)
• Degree audit and credit evaluation data (Domain 2)

PRIMARY ROLE OWNER

Provost / VP Academic Affairs

EVIDENCE

Arizona State University's eAdvisor improved four-year graduation rates from 33.2% (2006 cohort) to 44.8% (2009 cohort) — 2,024 additional graduates, an 11.6-point gain, with no change in entering SAT scores or class rank. Students on the correct course path rose from 22% to 91% in three years.

REFERENCES

• EvoLLLution, 2014
• ASU News, 2015

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Faculty lack real-time visibility into which students are disengaging within a course. At-risk students continue undetected until mid-term grades surface the problem — by which point the window for an effective intervention has often already closed.

HOW IT WORKS

Within a single course, a model combines a student's grades-to-date, demographic and academic-history data, and LMS click/login activity to compute a red/amber/green risk signal, which is shown directly to the student and routed to the instructor mid-semester. Unlike institution-wide alert systems, this operates at the course level: a real-time signal specific to one student’s current engagement in one class, not their historical risk profile across the institution.

DATA PREREQUISITES

• Student academic history and demographic records (Domain 1)
• LMS click, login, and engagement data (Domain 3 – Faculty & Teaching)
• Grades-to-date and assignment completion within the current course (Domain 3)

PRIMARY ROLE OWNER

Provost / VP Student Success

EVIDENCE

Purdue University's Course Signals showed 87.4% vs. 69.4% four-year retention for users vs. non-users in Purdue's own published data. Starting in 2013, these claims were publicly contested — critics documented in EDUCAUSE Review argued the effect was better explained by reverse causality: struggling students who were about to withdraw simply took fewer Signals-enabled courses. Independent causal replication does not exist. Independent researchers have since disputed the causal claim, arguing the effect may reflect struggling students dropping Signals-enabled courses rather than the tool improving outcomes.

REFERENCES

• EDUCAUSE, 2012
• EDUCAUSE, 2013

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Institutions invest in many programs without knowing which actually move the needle for which students. Resources flow to programs that correlate with already-successful students opting in, not to programs that causally improve outcomes. The distinction matters for every budget decision student-success leadership makes.

HOW IT WORKS

Propensity-matched scoring compares retention and persistence outcomes for students who did vs. did not participate in a specific intervention — coaching, tutoring, advising — controlling for self-selection bias. Students who participated are matched against a statistically similar group of non-participants, isolating the program's causal contribution from the effect of motivated students simply opting in to available support.

DATA PREREQUISITES

• Student enrollment, retention, and persistence records (Domain 1)
• Intervention participation records: coaching, tutoring, advising (Domain 1 / Domain 2)
• Academic program and course completion data (Domain 2)
• Financial aid and fee records (Domain 8)

PRIMARY ROLE OWNER

Provost / VP Student Success

EVIDENCE

University of Texas at San Antonio used Civitas Learning's initiative analysis to evaluate its Student Success Coaching program, achieving a 16-percentage-point increase in first-year retention and a 14-point increase in six-year graduation rate from 2012 to 2022.

REFERENCES

• Civitas Learning, 2022
• Civitas Learning, 2024

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Admitted students fail to complete required enrollment tasks over the summer: FAFSA verification, immunization records, housing deposits, registration. Staff cannot personally contact thousands of admits at the required pace. Tasks go incomplete and students melt away before the semester starts, forfeiting a seat the institution has already reserved.

HOW IT WORKS

A text-message-based conversational AI answers admits’ questions and proactively nudges them through required enrollment tasks at a volume no human team could match. Of 50,000+ student messages in GSU’s trial, less than 1% required escalation to staff. Effectiveness was established via randomized controlled trial: half of admitted students were randomly assigned to chatbot outreach; the other half received standard outreach only, allowing direct outcome comparison.

DATA PREREQUISITES

• Admitted student enrollment status and task-completion records (Domain 1)
• Financial aid document and verification status (Domain 11 – Compliance)

PRIMARY ROLE OWNER

VP Enrollment Management

EVIDENCE

Georgia State University’s “Pounce” chatbot (Mainstay/AdmitHub): RCT found a 21.4% reduction in summer melt for students who committed by the priority deadline; enrollment increase reported as 3.3-3.9% across two publications of the same study. GSU estimated the bot handled the messaging volume equivalent to ~10 full-time staff. A subsequent RCT found a 50% decrease in withdrawals due to outstanding balances.

REFERENCES

• Mainstay, 2022
• Brookings Institution, 2022
• Campus Technology, 2017

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Students who are academically on track drop out over small unpaid balances — amounts modest relative to the lifetime cost of losing a student permanently. Financial aid offices cannot identify and reach these students at scale without an automated trigger. The institution pays the full cost of attrition while the fix is a fraction of that cost.

HOW IT WORKS

Students facing a small unpaid balance who are otherwise academically on track are automatically identified from enrollment and balance data — no application required. Their balance is cleared up to a cap of $2,500, typically in exchange for a required action such as meeting a financial counsellor. This is a targeting-and-disbursement automation built on existing enrollment and balance data, not a predictive risk model.

DATA PREREQUISITES

• Student enrollment and academic standing records (Domain 1)
• Unpaid balance and financial aid disbursement data (Domain 8 – Finance)

PRIMARY ROLE OWNER

VP Student Affairs / CFO

EVIDENCE

Georgia State Panther Retention Grants: average grant ~$900, balance cap $2,500. 82-88% of recipients graduate or remain enrolled 12 months later. More than 19,000 awards issued by 2023. Program independently evaluated by Ithaka S+R (2022), confirming positive effects on retention and graduation. Each grant is disbursed automatically, with no student application required.

REFERENCES

• "Panther Retention Grants,” Georgia State University
• Ithaka S+R, 2022
• Inside Higher Ed, 2022

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Course schedules are rolled forward from last year’s section counts rather than built against actual demand. Bottlenecked, oversubscribed courses delay students’ time to degree, while other sections run under-enrolled, wasting seats, faculty time, and classroom space.

HOW IT WORKS

Historical enrollment patterns and declared program pathways are run through a forecasting model that predicts how many seats each course section will need next term. The registrar builds the schedule to match that projected demand instead of rolling forward last year’s section count — reducing both empty seats and the bottlenecked, oversubscribed courses that delay graduation.

DATA PREREQUISITES

• Historical course enrollment and section-fill data (Domain 2)
• Declared program/major pathways and demand signals (Domain 1 – Student)
• Classroom and facility capacity data (Domain 9 – Facilities)

PRIMARY ROLE OWNER

Registrar / VP Academic Affairs

EVIDENCE

Ad Astra's platform is used by 500+ institutions. Germanna Community College: Degree Velocity rose 15% (41% to 47%), with a 2% increase in credits earned per semester, confirmed against Ad Astra's own case study. Bossier Parish: 10.9% enrollment growth confirmed.

REFERENCES

• "Germanna Community College: Creating Scheduling Efficiency,“ Ad Astra
• "BPCC Case Study: Boosting Outcomes with Smart Plans,” Ad Astra

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Curriculum structure, prerequisite chains, and bottleneck courses silently delay student completion. These structural problems are invisible in grade data alone and are rarely surfaced until a program review, by which point years of cohort delay have already accumulated.

HOW IT WORKS

Every undergraduate degree plan is modeled as a directed graph of courses and prerequisites. Software computes structural metrics, centrality (how much of a choke point a course is), blocking factor, and delay factor, to mathematically identify which specific courses are extending time-to-degree, independent of any faculty member’s opinion about the curriculum.

DATA PREREQUISITES

• Degree plan and curriculum maps with prerequisite definitions (Domain 2)
• Historical course completion and grade data by cohort (Domain 1 – Student)
• Student registration sequence and actual course-taking patterns (Domain 1 – Student)

PRIMARY ROLE OWNER

Provost / VP Academic Affairs

EVIDENCE

University of Texas at Dallas is one of ~30 universities in the Curricular Analytics Project (Sept. 2021to Aug. 2024), computing centrality, blocking factor, delay factor, and complexity scores for all undergraduate programs. UT Dallas’s own program page describes the project as “validating the relationship between curricular structure and student outcomes.” The outcome metrics remain structural and qualitative rather than quantified to date.

REFERENCES

• UT Dallas Curricular Analytics Project

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

High DFW (D grade, Fail, Withdrawal) rates in gateway math and science courses block student progression, disproportionately affecting underprepared students. Fixed-pace lecture formats serve neither the students who are already ahead nor those who are behind — everyone moves at the same speed regardless of individual knowledge gaps.

HOW IT WORKS

Each student works through material on a platform that opens with a diagnostic knowledge check, places them at the point in the curriculum matching what they already know, and continuously reassesses and resequences material to their actual gaps. At ASU, the courseware fully replaced the textbook, exams, and lectures, not a supplement to existing instruction but a redesign of the course itself.

DATA PREREQUISITES

• Individual student learning-activity and assessment data within the courseware (Domain 3)
• Student academic history and course placement records (Domain 1 – Student)
• Course completion and grade outcomes (Domain 1 – Student / Domain 2)

PRIMARY ROLE OWNER

Provost / VP Academic Affairs

EVIDENCE

Arizona State University redesigned College Algebra with ALEKS adaptive courseware, winning the 2018 OLC Digital Learning Innovation Award ($100,000). Success rates rose from 55% (AY15-16) to 75% (AY16-17) to 79% (AY17-18) — a 20-point gain in two years, with 800+ additional students passing in year one. Underprepared students improved by 28 points. A later “stretch semester” option brought rates to 84%. Figures confirmed against OLC’s own award citation and Inside Higher Ed’s reporting.

REFERENCES

• Online Learning Consortium, 2018
• Inside Higher Ed, 2019
• Campus Technology, 2018

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?

INSTITUTIONAL PROBLEM

Departments cannot see teaching productivity against instructional cost. Resources flow to departments based on history and advocacy rather than actual section demand, producing both over-resourced units running under-enrolled sections and high-demand units short of capacity — a mismatch invisible in traditional budget processes.

HOW IT WORKS

Course-load, section-size, and instructional-cost data are compared at the department level against teaching output. The analysis shows deans where capacity is mismatched to demand. For example, a small department running many under-enrolled sections alongside a high-demand department that is short on sections, supporting decisions to reallocate sections or staffing rather than apply blanket budget cuts.

DATA PREREQUISITES

• Faculty course assignment and teaching-load records (Domain 3)
• Section enrollment and fill-rate data (Domain 2 – Academic Programs)
• Instructional cost and compensation data (Domain 8 – Finance)

PRIMARY ROLE OWNER

Provost / Dean / CFO

EVIDENCE

EAB publishes analyses of faculty workload, instructional capacity, and teaching productivity as financial-health levers, used at member institutions to reallocate sections and reduce part-time instructional cost. Specific institutional outcomes are typically confidential. This use case is primarily a financial diagnostic rather than a student-outcome driver; its value is in revealing cost mismatches that aggregate budget reports mask.

REFERENCES

• EAB Research and Insights

BEFORE YOU SCORE

1. Which of your active value drivers does this use case most directly serve?
2. Do you have, or can you reasonably obtain, the data prerequisites listed above?