Industry

    Pediatric therapy video without filming children

    ABA, speech and OT clinics cannot build a video library on real clients. How to use proxy avatars for parent onboarding and session-preview content instead.

    Versely Team9 min read

    Every pediatric therapy clinic that has tried to build a real video library hits the same wall in the same week. A recognizable child inside your clinic is protected health information, because the image discloses that this child receives treatment from you. Marketing use of that image needs a written authorization from the guardian that specifically covers video and the channels you intend to publish on. A verbal yes at pickup is not it.

    Then the authorization turns out to be revocable, so the asset has a half-life tied to the family's relationship with you. Then the child grows, and the eval video you shot with a four-year-old looks stale by the time they are six. The math never works out at volume. One video, maybe. Twenty-four videos a year, never.

    Which is why the clinics that actually ship consistent parent-facing content stopped trying to film clients at all and moved the child out of frame entirely.

    What "out of frame" actually means

    There are two versions of this, and only one of them survives contact with parents.

    The version that fails is generating photoreal children. Many image and video models restrict or outright refuse photoreal depictions of minors, and the outputs that do get through land in an uncanny register that parents notice instantly. This audience spends its whole life looking closely at children's faces. They will clock a synthetic one before the second cut, and the thing they will conclude is that you were willing to fake a client.

    The version that works is a stylized proxy plus real adults. One illustrated or clearly non-photoreal child stands in across the whole series. Your actual clinician appears on camera as themselves. The room, the materials, and the equipment are real or generated as environment b-roll with nobody in them. Nothing in the frame implies "this is a patient of ours," because nothing in the frame is a patient.

    That distinction is worth making explicit in your own internal brief, because it is also the line your compliance reviewer cares about. A stylized proxy is set dressing. A photoreal child is an implied person, and an implied person in a clinical setting is an implied client.

    Craft rules for proxy-avatar builds

    Six things separate a proxy series that reads as thoughtful from one that reads as cheap.

    1. Frame the materials, not the face. Hands on a token board. Articulation cards fanned out. A sensory bin, a therapy swing, a weighted blanket, a set of blocks mid-stack. Pediatric therapy is unusually legible through its objects, and parents recognize every one of them. Most of a session-preview video can run on materials alone.
    2. Use over-the-shoulder and back-of-head framing for the proxy. The child's presence is established by scale and posture. You do not need a face to communicate that a kid is sitting at the table.
    3. Keep the clinician real. A credentialed BCBA, SLP, or OT on camera is the trust anchor and the only human the video needs to individuate. Build them from a consented reference clip through the AI avatar generator so a script revision is a re-render rather than a re-shoot on a clinic day.
    4. Lock the proxy across the series. One proxy child, one clinician, one room, every video. Parents watching three of your videos should feel like they are looking at the same place. This is a character consistency problem, so build a reference set once and reuse it rather than re-prompting from scratch per video.
    5. Never use captured room audio. Background chatter in a therapy room is identifiable, and a name spoken off-camera is a disclosure. Generate the voiceover from the approved script instead. This also means the same footage can be re-narrated for Spanish or for a different age group without touching the picture.
    6. Say what the proxy is. A single line in the description, and a super on the first appearance: illustrated stand-in, not a client. It costs you nothing and it answers the question every parent is already forming.

    Build one: the parent onboarding series

    This is the higher-value library and almost nobody makes it. Pediatric therapy clinics tend to be waitlist-constrained rather than lead-constrained, so the marketing job is not filling the top of the funnel. It is reducing the phone burden, cutting eval no-shows, and getting families through authorization without four rounds of email.

    Six videos, 60 to 90 seconds each, that pay for themselves in front-desk time:

    • What happens at the evaluation. Parking through handoff. Who will be in the room, how long it runs, what you will and will not know that day.
    • What "prior authorization" means and why it takes weeks. A recurring intake frustration, and one easily defused by a clinician explaining it in plain language.
    • What to bring, and what not to. Comfort item yes, full lunch no, the IEP if there is one.
    • Your cancellation and attendance policy. Deliver it as video and the awkward conversation stops happening at the desk.
    • What progress actually looks like. Sets a realistic timeline without promising an outcome, which is both honest and the thing that keeps families enrolled past month three.
    • How to talk to your child about coming here. The one parents tend to forward, usually to grandparents.

    Script these as clinician-to-camera with proxy b-roll under the middle third. The explainer video generator handles the structure; environment shots come from the b-roll generator with prompts that specify an empty room. "Pediatric therapy room, sensory swing, soft afternoon light, no people, five seconds" is the shape of prompt you want.

    Build two: the session preview

    Parents on a waitlist have one recurring question: what actually happens in there. Clinics answer it badly, usually with a stock photo of a smiling child and a paragraph about evidence-based practice.

    The 90-second version that works:

    1. Ten seconds of clinician on camera. Name, credential, discipline, and what this specific video shows.
    2. A wide of the room, empty. Establishes scale and cleanliness. Parents are assessing the space as much as the method.
    3. Forty seconds of the session in beats, materials-first. Greeting and transition. First activity and why it looks like play. The reinforcement or cueing moment. A break. A second activity. Voiceover carries the clinical reasoning while the picture stays on hands and objects.
    4. Fifteen seconds on what the parent does. Where you sit, whether you watch, what the handoff at the end sounds like.
    5. A close that sets expectations, not outcomes. What you will hear from us after session one, and when.

    Cut it on a single timeline in the video editor so the variants for speech, OT, and ABA share structure and swap only the middle section. The 480p preview pass is free with a short per-user cooldown, which is enough for your clinical director to sign off on pacing before you spend credits on a final export. That export is charged once no matter how many clips are on the timeline.

    Burn in captions with the caption generator. Parents watch this at 10pm with a sleeping kid in the next room, and a waiting-room screen has no audio at all.

    Where the rules differ

    Two situations change the analysis, and both are common in this vertical.

    If you deliver services inside a school, the records framework may be the school's rather than yours, and the permissions you need are not the ones your intake packet collects. Ask the district before a camera or a phone comes out on that campus, and assume the answer is stricter than you expect.

    If you run group sessions or social skills groups, incidental capture is the risk. Every child in the frame is a separate authorization, and one missing signature makes the whole clip unusable. Proxy builds sidestep this entirely, which is the practical reason group programs are the hardest thing in the category to market with real footage and the easiest to market with stand-ins.

    For the wider set of consent and disclosure questions that come with generated presenters, stock avatar, digital twin, or generated character walks the three options and what each one obliges you to say. The adjacent clinical verticals are covered in AI video for pediatricians and AI video for therapists and mental health practices.

    FAQ

    Can we ever use a real client with a signed authorization?

    Yes, and one flagship family story with proper written authorization is worth having. Treat it as a one-off with a defined review date, not as the model for your library. The authorization is revocable, so build the asset in a way you can pull and replace without gutting a series. Keep the plan for what happens if it is revoked written down before you publish.

    Does an illustrated proxy need a disclosure?

    Practically, yes, and it is cheap. A super on first appearance and a line in the description. The point is not a legal formality, it is that a parent who wonders whether that was a real client and cannot tell will assume the worse answer. Answering before they ask converts better than the ambiguity ever would.

    What about photos parents post and tag us in?

    A parent posting their own child is their disclosure to make. Resharing it from your clinic account is yours, and it is a different act. Get written permission for the reshare specifically, and do not turn it into an ad without an authorization that says so.

    Does the same approach work for adult therapy?

    The consent problem is milder because adults can authorize for themselves, but the revocability and the aging-out problems remain. The proxy method still tends to win on cost and durability once you are shipping more than a handful of videos a year.