Your next implementation, dashboard, or AI agent will inherit every definition your offices disagree on. Three short exercises show you where the gaps are.
BY AMY MALTAGLIATI AND BROOKE MOORE
Picture your president’s cabinet meeting. Enrollment is first on the agenda. The VP of enrollment reports one number for enrolled students, the registrar another, and the provost a third. Someone asks which is right. Nobody can say for certain because everyone is using a different definition of “enrolled”. This exact conversation has happened countless times with our partners over the years. While it has always been an issue worthy of addressing, AI’s integration into business processes and decision-making makes it more urgent than ever. Where humans understand nuance and move on while accepting the different definitions, AI chooses one winner, codifies it, and incorporates the “winning” definition into every future decision. The consequences can be significant.
Data governance in higher education comes down to three tests. Aligned: Do your systems agree on what words mean? Owned: does a named person have authority to rule on each definition? Ready: can a system reach the answer, or does it live in someone’s head? Together, these three quick but critical tests show where your CRM data governance breaks and provide a framework for building stronger data integrity on your campus.
New CRM tools inherit the data problems you bring to them
The team that implements a higher education CRM usually runs it afterward, and it is stretched thin before the project starts. Cleaning up legacy data, building standards, and defining who-owns-what decisions (the unglamorous stuff) are often the first items deferred on CRM implementation plans.
“We’ll clean it up later” is said in good faith at nearly every kickoff planning session. Too often, “later” never comes. The team moves on to new projects and people change roles. The same duplicate-record and field definition debates return every year. Nothing restarts the conversation until a new project comes around, or a new person leaves, or a new initiative is identified.
“A new solution, a tool, a bolt-on AI — they’re not going to fix your governance problem, but they will inherit it, and then it’ll make it permanent in your new field structures,” says Brooke Moore, director, delivery excellence at Human Capital Education.
If you’re moving platforms, settle definitions and owners before anyone builds a field. It is never a bad time to harden definitions and ownership, but it is critically important when launching a new Salesforce CRM implementation project, or undertaking a significant CRM optimization effort. It is the first conversation to have with any higher education CRM implementation partner.
Aligned data means your systems agree on what words mean
Aligned data means the systems you already run, including your student information system, your CRM, and every connected tool. Use the same definition for the same word and point to one agreed source of truth. The truth is that nobody decides to disagree on definitions. Misalignment builds up one integration and one department at a time, until “enrolled” or “applicant” means something different in each office’s reports.
Naming standards reduce unintended records duplication
CRM data quality quietly erodes without anyone noticing every time a record is created with poorly defined naming standards. Your foundation team tracks one corporate partner relationship, and the partner’s name appears five ways across three opportunity records. The CRM can’t merge them, can’t show the relationship’s true value, and can’t stop two staff members from contacting the same person in the same week.
“A naming standard takes an afternoon to write down, but unwinding all of the duplicate records after the reports are built on top of them can take months,” Moore says. Writing field definitions and naming standards into the CRM itself is core higher education CRM optimization work.
Two legitimate definitions need two labeled fields
Alignment doesn’t require one office to win. When a grant-funded program has a real reason to define a term differently, give each definition its own named field with its own help text. Amy Maltagliati, senior director and Salesforce practice lead at Human Capital, recommends “you don’t want the same field to mean two different things to two different offices.” If the term matters enough to have two different but clear definitions, create two clearly distinct fields with clearly distinctive labels.
When we asked recent attendees of our webinar on Building a Data Foundation Your Next Salesforce Investment Can Depend On how closely their registrar’s and advising office lists of active students would match, no one chose “identical.” The most common answer was “different enough that it would matter.” That is concerning and requires an immediate solution.
Owned data has a named person with authority to rule on it
Data governance in higher education is the set of decisions, and the accountability behind them, that determine how your institution’s data is created, used, and controlled. At Human Capital, we group those decisions into five areas. We call them the High Five:
- Who can access data?
- How accurate is the data?
- How is student privacy protected?
- How secure is the data?
- How long is it kept before it is retired?
Data stewards, custodians, and sponsors each own a different decision
Higher education data governance succeeds or stalls on a single question: who decides? “Governance isn’t a policy binder,” Moore says. “It’s a set of decisions that someone actually has to make.” Decisions without named owners aren’t made — they’re re-litigated, every time they resurface. The five roles below define who holds which decision, and who must sign before it changes.
A data steward owns data quality for one domain, such as admissions, the registrar, or advancement. A custodian manages the systems, access, and compliance around that data. An executive sponsor keeps governance alive past the first competing priority, and technical builders and analysts make sure no decision is made without the people who implement it.
Anyone who might be asked for a number holds a stake in data integrity. Defining the stakeholder roles doesn’t require the development of new jobs, but it does require that specific and clearly defined roles belong to specific people.
Most institutions assign responsibility without authority
We asked our webinar attendees who makes the call when a number in a leadership report turns out to be wrong.
| Who decides when a leadership number is wrong? | Share of attendees |
| An office owns it, but no one person does | 38% |
| A specific person (I could name them) | 31% |
| It depends on the number | 15% |
| We'd all look at each other | 15% |
When an entire office owns a number, rather than an individual, someone has custody but no one can rule. Naming a person to each role is a critical start. But, the true test of governance is a simple question: Could that person actually overrule a VP? If you say yes, you have true governance. If you say no, you just have a very patient colleague.
Effective ownership rules also have to survive staff turnover: “the person who made that decision may not be there anymore,” Maltagliati says. Name a backup for vacations, too. When a steward or administrator seat sits empty, a dedicated CRM consultant can keep the work moving while you fill it.
A governance council needs authority and a queue of decisions
A data governance council is where owners settle cross-department disputes once, approve changes to shared definitions before they go live, and prioritize the reporting backlog. Keep it to five to eight people. It fails without authority or without decisions to make, and it needs four things:
- A one-page charter stating what the council has authority over
- A standing monthly meeting (quarterly is too slow, and requests pile up)
- A documented change-request process for proposing, reviewing, and communicating changes
- A source-of-truth log of every approved definition and its owner
A group of people that can only read data and information without the authority to stop a field change is just a book club. Ensure your governance council has true decision-making authority that spans departments on campus.
Ready data is both trustworthy and reachable
Ready data is trustworthy and sits where a reporting tool, a predictive model, or an AI agent can reach it. Accurate data in a PDF, a spreadsheet, a notes field, or a long-tenured employee’s memory fails the test. Harden your data systems and formatting to ensure your data is both trustworthy and reachable.
Free-text notes need one structured field beside them
Advising and call notes hold rich information that an AI agent simply can’t use. It’s not useless. It’s just unreachable. The solution doesn’t require you to rework years of history, but it does need to be addressed. The solution is to capture at least one structured element with each new entry, such as a date, a category, or a resolution flag, so a system can find those advisor call notes.
AI agents remove the person who used to catch the error
Definition gaps used to be caught by a person before a report reached the cabinet. Ask for a list of active students, and the person building it asks which field and how far back. Automated reports, portals, and AI agents remove that person, so a single definition of an “engaged” student becomes a rule applied consistently to every student, every day.
To illustrate this cost, consider a student with duplicate records across campus. Two offices create two separate records for one student and identity resolution processes fail to match the records. When your new AI agent is asked whether a student submitted a form, it pulls the wrong record and reports the form as missing. Suddenly, a student who has followed every process may face consequences for an error they didn’t make. The same gap reaches every enrollment analytics dashboard built on those records.
In Salesforce’s State of Data and Analytics report, 89% of data and analytics leaders with AI in production said they had experienced inaccurate or misleading AI outputs, and 89% called a strong data foundation the most critical factor for successful AI. Data preparation belongs at the front of any AI strategy and execution for higher education.
Three short exercises find the gaps in 90 minutes
You don’t need to survey the whole campus over several weeks to find your data integrity gaps. Here are three quick exercises that each fit inside a 30-minute meeting. At the end of each exercise, you’ll have a one-page document that clearly defines your current data governance standing. We recommend including representatives from your IT and CRM team in every exercise, so they hear the gaps firsthand from the end users.
| Exercise | What it tests | Who attends | Time | What you walk out with |
| Five-Term Audit | Aligned | One person from each office that uses the terms, plus your CRM or systems administrator | 20 minutes | A one-page table, one row per term, mismatches circled |
| Ownership Gap Test | Owned | The people who run each data domain day to day, your CRM administrator, and your executive sponsor | Up to 3 minutes per domain, 4–5 domains | A one-page table, one row per domain; every blank cell is a finding |
| Query Readiness Test | Ready | The people who would build or manage the tool, plus the functional leader who owns the question | 20 minutes | A pass/fail scorecard with one next step |
Step 1: Run the Five-Term Audit to test alignment
The Five-Term Audit produces a one-page table showing which shared terms your offices define differently. Pick five terms you report on often, such as “enrolled,” “active,” “at risk,” and “major gift.” Write definitions in silence for five minutes, read them aloud for 10, and circle every mismatch in the last five. Send the table to whoever owns CRM reporting, and update your data dictionary once the definitions are settled.
Step 2: Run the Ownership Gap Test to find who can rule
The Ownership Gap Test produces a one-page record of who owns each data domain, with the gaps visible. List four or five domains, such as student records, donor records, course data, and financial aid. For each, ask: Who owns this today? Who can change a definition? What happens before a change goes live? Record silence as a finding, and send the page to whoever holds authority over your systems.
Step 3: Run the Query Readiness Test before you build
The Query Readiness Test produces a pass/fail scorecard for one real question a system would need to answer, such as “Which admitted students are at risk of not enrolling?” Run four checks. Is there a clean, agreed answer? Is the data reachable? Is it current enough to trust? Is a named person watching for drift? Score it, name one next step, and take the scorecard to whoever owns your data or AI roadmap as a go/no-go.
Find the gaps during design, not in front of your cabinet
Under pressure to launch an AI pilot? You don’t have to stop. Run the Query Readiness Test on its question first. If it fails, you’ve just found out something during design instead of during a demo in front of your cabinet. This quick exercise helps you avoid the trap in which AI pilots show promise but outcomes stay flat.
Human Capital's technology audits go deeper
When the exercises surface more gaps than your team can close, Human Capital Education’s higher education CRM audit reviews your configuration, data, integrations, and processes, then delivers a prioritized plan of quick wins and structural fixes. If you’re staying on your current platform, a CRM optimization consultant can turn that plan into rebuilt fields, checklists, and reports. Contact our higer education CRM and Technology experts to explore how we can help your institution build better data governance processes and stronger technology solutions.
Frequently Asked Questions
We already have a data dictionary. Is that enough?
A dictionary records decisions; the Five-Term Audit is a meeting, and its value is the moment two people read conflicting definitions aloud. Hold the meeting, then update the dictionary and revisit it before every new tool or implementation.
What's the difference between a data steward and a data custodian?
The steward decides what a field means for one domain and answers for its quality. The custodian controls the system around it: who can access it, how it’s secured, and whether changes meet compliance requirements.
Our governance committee stalled. How do we restart it?
Before you reconvene anyone, name one person who must sign off on changes to shared definitions. Then bring the Five-Term Audit table as the first agenda, so the group starts with real decisions in its queue.
Do we need to pause our AI pilot to fix our data?
No. Trace the pilot’s own question end to end with the Query Readiness Test before configuration starts. If it passes, you’ve spent half an hour; a failed check becomes a specific fix list while changes are still cheap.
About the authors
Brooke Moore
Director, Delivery Excellence & Practice Enablement, Human Capital Education
Brooke Moore is an education strategist with over 18 years in higher education and holds an Ed.D. in Higher Education Leadership and Administration. Before consulting, she served as Assistant Vice President for Enrollment Management, leading cross-functional strategies to enhance access, affordability, and student support. Brooke bridges technology and campus operations, ensuring solutions align with institutional goals.
Amy Maltagliati
Senior Director, Salesforce Practice Lead, Human Capital Education
Senior Director, Salesforce Practice Lead, Human Capital Education
Amy Maltagliati is a higher education strategist with 15+ years in admissions, advising, and enrollment management across community colleges, online institutions, and R1 universities. A Salesforce-certified professional, Lean Black Belt, and Certified Agile Scrum Master. She partners with leaders to translate strategic priorities into scalable CRM solutions that drive enrollment and student success.