Video for yield season and summer melt
Between the deposit and move-in day the job is reassurance, not recruitment. A personalised-at-scale video plan that needs no new footage and no new consents.
An admitted student who paid a deposit has already been convinced. They toured, they read the viewbook, they told their family. Then they stop enrolling — not because they changed their mind about the institution, but because a housing portal timed out, a financial aid verification request landed in a spam folder, or nobody at home could explain what a bursar is.
That gap between deposit and matriculation is summer melt, and video is unusually good at closing it. But the video that closes it looks nothing like the video that produced the deposit. Recruitment video sells a future. Yield video removes friction from a checklist. If your summer sends are recut recruitment footage with a new title card, you are answering a question the student stopped asking in April.
Yield season is a task-completion problem
List the things a deposited student has to do between May and August. At most institutions it looks something like this:
- Confirm the enrolment deposit and accept the offer
- Complete financial aid verification if selected, and understand the award letter
- Submit immunisation and health records
- Apply for housing, complete the roommate questionnaire, sign the licence
- Take placement tests in maths and language
- Register for an orientation session and, at many schools, register for courses
- Set up institutional accounts, email, two-factor, student ID photo
- Read and pay the first bill, or set up a payment plan
- Handle move-in logistics — date, time slot, parking, what fits in the room
Nine tasks, most of them administrative, each with a deadline, none of them exciting. Research on melt has consistently focused on students who do not have someone at home who has already navigated this — first-generation students, low-income students, students whose counsellor relationship ended the day they graduated. The friction is not motivational. It is procedural.
Which is good news, because procedural friction is the easiest thing in the world to film. You do not need an emotional arc. You need somebody to point at the right button.
What you can build without new footage or new consents
The constraint that stops most enrolment teams is consent, though not quite for the reason usually given. FERPA is narrower here than marketing teams assume: the regulation lists a photograph among the items an institution may designate as directory information and release under a public notice-and-opt-out process, so a flat "FERPA forbids it" does not hold on its own. What actually forces a signed release is institutional policy, right-of-publicity law, and nobody wanting to argue after the fact over whether a particular frame was an education record. The paperwork problem is real either way: releases gathered for last year's recruitment shoot may not cover a summer campaign, may not name the students you now want to feature, and certainly do not cover students you have not met yet.
Work around it rather than through it. Every asset below is available without a new shoot and without a new release:
| Ingredient | Source | Consent status |
|---|---|---|
| Campus exteriors, quads, walkways | Filmed empty, early morning | No individuals in frame |
| Building interiors, labs, residence hall model room | Filmed empty | No individuals in frame |
| Screen recordings of the actual portals | Captured from a test account | No student record shown |
| Staff and faculty explaining their own process | Filmed, staff releases | Standard employment release |
| Seasonal exteriors and weather you did not shoot | Generated | Must depict the real campus honestly |
| Motion graphics, deadline cards, checklists | Typed overlays over generated backgrounds | No record data in the picture |
| Narration | Voiceover | Synthesised narrator is fine |
The screen recordings are the single highest-value asset in that table and almost nobody makes them. A 40-second capture of the housing portal — this menu, this button, this is where the confirmation email lands — outperforms a drone shot of the campus by a distance during yield season.
There is a nuance worth stating plainly, because teams over-correct on it. A personalised video sent by the institution to the admitted student themselves is not a disclosure of that student's record to a third party. The exposure sits in publishing a recognisable student to an audience, not in addressing one privately. So personalising the send is fine. Featuring a real student's face in it is the part that needs paperwork.
One master, four swap layers
Build each send as a single timeline with named swap slots, then vary the slots. Saving a video edit as a reusable draft is what makes this a fifteen-minute job per segment rather than a fresh project.
| Layer | What varies | How it is filled |
|---|---|---|
| Picture | Nothing | Shared master — campus, portal capture, staff |
| Text overlay | Name, college, deadline date, advisor, orientation session | Typed overlay driven by a merge sheet |
| Voice | Same script with the segment noun swapped | Regenerated line only, not the whole track |
| Tail card | Segment-specific next action and contact | Typed overlay |
Two rules keep this from breaking. First, personalised values go into text and voice, never into generated pixels — a model rendering a student's name will misspell it, and a misspelled name in a yield email is worse than no name. Second, segment on something that changes the advice, not on something that flatters. "Engineering" changes which placement test matters. "Out of state" changes the move-in section. Hometown does not change anything and reads as surveillance.
Segments that earn their keep: college or school, on-campus versus commuter, in-state versus out-of-state versus international, and honours or scholarship cohorts with extra requirements. Four dimensions, and most institutions can cover the real variance with eight to twelve variants per send. That is a batch generation job, not a production schedule.
The seven-send calendar
| Timing | Trigger | The video's one job |
|---|---|---|
| Deposit + 48 hours | Deposit recorded | Show the whole summer on one screen so nothing is a surprise |
| Mid-May | Housing application opens | Walk the housing portal, roommate form, licence signature |
| Early June | Verification selection posted | Explain what verification is, what to send, who to call |
| Orientation − 7 days | Session registration | What happens that day, what to bring, where to park |
| Mid-July | Advising window opens | How registration works and what to do if a class is full |
| Early August | First bill issued | How to read the bill, payment plan options, the due date |
| Move-in − 3 days | Move-in slot assigned | Time slot, route, what is open, what to bring |
Keep each to 45–90 seconds. Ship two cuts of every send: a vertical version for SMS and social, and a still frame with a play affordance for the email body, because most institutional email clients will not autoplay anything. Email marketing with generated video covers the delivery mechanics, and the personalisation logic is the same shape as account-based video personalisation on the B2B side.
For the two or three sends where a real student voice genuinely helps — orientation, move-in — record current students you already have signed releases for, and record them as audio over campus b-roll rather than on camera. Audio-only consent is easier to obtain, easier to scope, and the asset survives the student graduating.
Measure task completion, not views
Views are the wrong metric here and they will make you feel good while melt continues. The measurable question is whether the send moved the task.
For each send, pull the completion rate for the underlying task at 7 days and 14 days after send, split by whether the student opened the video. Then compare against the same window in a prior cycle. You are looking for a shift in the completion curve, not for engagement.
Two things usually fall out of that analysis in the first year. One send does almost nothing and can be cut — usually the mid-July advising one, because advising already has a human chasing it. And one task has a completion cliff nobody had noticed, usually financial aid verification, which justifies three videos instead of one. Run the sends as a proper creative test rather than replacing the whole calendar on a hunch.
The wider university content programme can absorb these assets afterwards. Portal walkthroughs date slowly, and a summer library becomes the transfer-student and readmit library in the autumn with a new tail card.
FAQ
Do we need consent for a student who appears only in a wide campus shot?
The line is recognisability. A wide shot where individuals are not identifiable is a different case from a mid shot of one student walking. Institutions handle this differently and your general counsel sets the threshold, not the marketing team — but the practical production answer is to shoot campus empty at 6am and avoid the question entirely. It also looks better, because you control the light.
Can we generate a student for the parts where we have no footage?
For b-roll texture, with disclosure, yes. For anything that presents as a current student describing their experience, no. That is a fabricated testimonial regardless of how it was produced, and in a sector where trust is the product it is a bad trade. Generated material belongs in seasonal exteriors, motion graphics and abstract transitions — the places where nobody is being represented.
How short can these be before they stop working?
The portal walkthroughs can run to 90 seconds because the viewer has a task and wants the answer. The reassurance sends should not exceed 45. The test is whether every second is doing work: if a shot exists to set a mood rather than to show a step, cut it. Yield-season video is closer to a support article than to a brand film.
What do we do with these when the cycle ends?
Re-tail-card them for the next admitted cohort and reuse the picture wholesale. Portal captures need re-shooting whenever the interface changes, which is the only recurring cost. Everything else — the empty campus, the staff explanations, the deadline card design — carries forward, which means year two of this programme costs a fraction of year one. The same applies to the back-to-school window for institutions running both.