For colleges and universities
An admitted student needs one document verified, one aid question answered, one credit evaluated. In July the office handling that is at its thinnest and its busiest at the same time. The student emails, waits, emails again, and quietly stops. Nobody records a loss, because nothing happened.
A student is admitted in the spring and enrols in the autumn, and everything that determines whether that actually happens occurs in between — in the weeks when the offices responsible are running on reduced staff, working through a verification backlog, and answering the same twelve questions individually, hundreds of times.
The student experience of that period is a series of unanswered messages. They ask about an aid package and are told to check a portal they cannot log into. They upload a document and receive no confirmation that it was the right one. They call an office that is closed, leave a message, and are called back during a shift at the job they took to pay for the term.
First-generation students absorb this worst, and they are the students the institution most wants to keep. They are least likely to know that a request for verification is routine rather than a rejection, and least likely to have someone at home who can tell them.
Transfer students face a version with a longer clock. Credit evaluation determines whether the offer is worth accepting, and it frequently arrives after the student has had to decide. An institution can lose a transfer student to its own internal timeline without ever making a decision about them.
Then census day arrives, the number is what it is, and the analysis begins. It will identify characteristics of students who melted. It will not identify which of them were still reachable in the third week of July, because nothing recorded that.
And every office involved is genuinely working hard. This is not a motivation problem. It is that the work of moving one student forward is spread across offices that each see their own segment and none of which owns the student’s whole path.
An admitted student who stops progressing generates no event in any system, so the population that is still recoverable is indistinguishable from the population that has already chosen elsewhere.
This is the whole difficulty. Systems record steps completed: deposit paid, document received, registration submitted. Nothing records that a student who completed step three has not attempted step four for nineteen days. Absence produces no row, and absence is the entire signal.
The consequence is that outreach goes to everybody or to a segment defined by a characteristic — first-generation, out-of-state, a particular aid band — rather than to the students whose progress actually stalled. That is why it converts poorly and why staff distrust it: they know most of the list did not need the message.
Making the stall visible changes the shape of the work. Instead of a campaign, there is a working list ranked by how long a student has been stuck and what specifically is blocking them — a missing document, an unanswered question, an unevaluated credit, an unresolved hold. Those are four different conversations and only one of them is about persuasion.
It also settles an argument institutions have every year. When staff say they are overwhelmed in July, the response is usually a request for evidence. The decomposition of a student’s elapsed time — how long they waited, for what, at which office — is that evidence, and it is the input to a staffing decision that currently gets made on impressions.
Students visible as stalled while they are still reachable — measured by the count of admitted students with a known blocker and days-since-progress, against a baseline where that count could not be produced at all.
Time from a document arriving to a student being told it was accepted — measured by elapsed hours from submission to confirmation, before and after.
Transfer credit evaluation reaching students before they must decide — measured by days from application complete to evaluation delivered, against your own admit-decision deadline.
Repeat contacts about the same unanswered question — measured by contacts per student per issue, which is the cleanest proxy for whether the first answer landed.
What the summer workload actually consists of — measured by staff time on repeated routine questions versus time with students who need an advisor, sampled the same way in both periods.
Outreach precision — measured by messages sent to students with a confirmed stall, as a share of all outreach sent.
Whether melt was unreachable or merely unreached — measured by melted students who had a recorded, unresolved blocker — a distinction census analysis cannot currently draw.
any academic advising, any admission decision, any aid determination, and any judgement about a student’s standing. Nothing here advises a student on what to study, decides who is admitted, or determines eligibility. Those belong to your professionals and to federal and institutional rules, and a system that blurred that line would be creating a compliance problem rather than solving an enrolment one.
This reads the systems already holding student state, through their documented interfaces. Nothing migrates, and no second student record is created — a parallel record of a student’s status is a data-governance problem and it will be found during the next audit rather than the next term.
Where an interface does not exist for a step, that gap is reported as a gap. It is not worked around by automating a screen, because a screen-scraped workflow breaks silently at the next upgrade and in this operation it would break during the six weeks that matter.
Access follows your directory and your existing role definitions, so a student worker who leaves at the end of the summer loses access when your directory says so rather than when somebody remembers.
Education records carry obligations under federal law, and those obligations attach through your institution and its agreements — not through a certification, because there is no FERPA certification regime and no vendor holds one. Any vendor answering that question with a yes is describing their sales process rather than their compliance posture.
Where an outside party acts on your behalf with respect to education records, the arrangement has to satisfy your own registrar and general counsel on their terms. The documentation for that review exists before anything is connected, including exactly what is reachable and under what authority.
Access is scoped to the population and the step in operation. A team working the verification queue sees the students in that queue and the records that queue requires, not the wider student body, and that boundary is enforced where it cannot be forgotten during a busy week.
Operational access is not permission to train. Student data does not become material improving anything that serves another institution. Given what an admissions file contains, that boundary should be non-negotiable, and it is architectural rather than a policy sentence.
The registrar owns the education-record question and should be in the room at scoping rather than at go-live. The boundary that satisfies them is usually simple to state: access is scoped to one queue, every touch is recorded, and nothing determines a student’s standing.
The CIO’s question is about integration risk and identity, and it is answered the same way every other integration is: documented interfaces, your identity provider, your deprovisioning.
The question nobody schedules is the faculty and staff one — whether this replaces the human relationship students are told they will have. It does not, and the scope should show it rather than assert it: what gets delegated is the repeated routine question, and the effect on your advisors is that they spend the summer on students who need an advisor.
We do not claim certifications we do not hold. Where an obligation applies only under a specific agreement or data class, it is marked applicability-gated rather than presented as standing capability.
A single admitted population and a single step in their path — verification, aid acceptance, transfer credit, or registration — observed read-only through one cycle, with no student contacted.
Observation contacts nobody and changes nothing. It produces the stall profile for that step: how many students are waiting, how long, on what specifically, and how that distributes across the populations you are most worried about losing.
That profile answers a question institutions currently answer at census, when the year is finished. It is worth having on its own, and for some institutions it will point at something internal — an office that needs a person in July, a deadline set earlier than the process can serve — that no vendor needs to be involved in fixing.
If you continue, the first delegation is narrow: one repeated routine question, on one channel, with your approved language, a written authority grant, and advisory escalation defined before a single student is contacted.
They should, and the summer is precisely when they cannot, because the same staff are working a verification backlog and answering the same twelve questions hundreds of times. The delegation proposed first is that repetition — not the advising. The intended effect on your advisors is that a student who needs one gets one, instead of waiting behind three hundred people asking where to upload a form. If a step cannot be cleanly separated from advising, it stays with your staff.
The honest answer is that nobody is: there is no FERPA certification regime, and a vendor claiming one has told you something about their diligence. What exists is a written arrangement your registrar and general counsel review on their own terms, plus the scoped access model, the audit trail, and documentation of exactly what is reachable. That is the reviewable object, and it should be reviewed before anything is connected.
An observation scope is read access to one step, which is a materially smaller ask than a project — and a system implementation is one of the few situations where an independent baseline is unusually valuable, because you will want to know whether student progression got better or worse across the cutover. If read access during the implementation genuinely cannot be granted, that is a fair answer and the right time to revisit is after stabilisation.
A dashboard reports what already happened to somebody who can no longer act on it, and at census that is exactly the wrong moment. The test to apply is whether the output is a working list a named office acts on this week — students who are stalled, ranked by duration, with the specific blocker named. If after one cycle no office is working from it and no student was reached because of it, it did not work and you should say so rather than renew it.
Then it should not, and the scope should reflect that from the first conversation. Observation of a verification queue is read-only and determines nothing. If your institution concludes that no part of that step may be delegated, the measurement still stands on its own — knowing how long students wait in verification and what they wait for is useful to your aid director whether or not anything is ever handed over.