Media selection for healthcare: which platform, for whom, and why most hospitals get it backwards

There is now peer-reviewed evidence of exactly this. A 2024 study in Digital Health examined the social profiles of all 194 public hospitals in Italian regional capitals. Only 36.6% had a social media profile at all, and just 18.3% had a social media policy. Of the active ones, 82.3% published the identical post on Instagram and Facebook with no modification. Average engagement was 0.62% of potential users.

That is not a strategy failing. It is the absence of one — platforms treated as distribution pipes rather than distinct social contexts with different audiences, different norms, and different jobs.

This piece is an attempt at the alternative: which platform, for which audience, doing which job, and what the combinations should look like.

The frame: what a media choice actually signals

Before the platform-by-platform analysis, one idea that reorganises everything downstream.

Rory Sutherland, Vice Chairman of Ogilvy UK, has spent a career arguing that human decisions follow what he calls psycho-logic rather than narrow economic logic — and that one of its most powerful mechanisms is costly signalling. His argument, in brief: a signal is credible in proportion to how expensive it was to produce. Advertising works partly because it is visibly expensive, which demonstrates a company's confidence in its own future. He compares it to a racehorse owner betting heavily on his own horse.

His favourite illustration is the London black cab driver's Knowledge — years of study for a job that satellite navigation could technically do. The point is not the navigation. It is that someone who has spent four years qualifying is unlikely to throw it away cheating a passenger. Sutherland explicitly extends this to medicine: a decade of training and enormous cost function as a trust placebo, a guarantee of skin in the game.

Which produces the observation this whole article turns on.

A doctor already possesses the strongest costly signal available in any profession. Most healthcare social media actively degrades it.

A consultant with twenty years of subspecialty practice posting to a trending audio clip is not building reach. They are substituting a cheap signal — one anyone with a phone can produce in ninety seconds — for an expensive one that took twenty years. The format says low investment. The credential says enormous investment. The format wins, because it is what the viewer actually experiences.

Sutherland makes a second point that runs directly against conventional conversion practice: friction can be a trust signal. In a world of abundant free information, the ease of communication devalues it. Something that visibly cost the sender effort carries more weight. His example is two wedding invitations containing identical information — one emailed, one on embossed card. You infer entirely different things.

For clinical media this inverts the standard advice. The considered twenty-minute explanation, the diagram the surgeon drew themselves, the article with forty references — these are expensive signals. They are also, not coincidentally, the formats that answer engines cite and referring doctors forward.

Cheap formats produce reach. Expensive formats produce trust. In a category where the purchase decision is "will I let this person operate on my father," reach without trust is worth very little.

The three audiences most plans conflate

Healthcare media has three distinct audiences, and almost every plan we review treats them as one.

The patient — searching, frightened, often not for themselves. Reached at the moment of diagnosis, not the moment of scrolling.

The referrer — a GP or a colleague in an adjacent specialty deciding whether to send someone. For most specialist practices this audience delivers the majority of patients, pre-qualified, with trust already transferred.

The peer and the profession — colleagues, societies, KOLs, conference discourse. Not a patient acquisition channel. It is where reputation is manufactured, and reputation is what makes the other two work.

Each platform serves these unevenly. That mismatch, not posting frequency, is where healthcare media budgets are lost.

Platform by platform

Facebook — the community and the older patient

Audience: patients and families, skewing older. Chronic disease communities.

Facebook is the most-used platform by healthcare institutions. In the Italian study, 87% of socially active hospitals used it as their primary channel — the only figure here we would stand behind, because US hospital adoption numbers circulating online trace to aggregators that contradict themselves.

The genuine job it does: private groups. Physician survey data from Sermo and LiveWorld found 52% of doctors use Facebook specifically to join private medical groups. The same structure serves patients — chronic disease communities on Facebook are among the most durable patient support infrastructure that exists, and for conditions with long management arcs they are where families actually talk.

Where it fails: as a broadcast channel for a hospital. Posting event announcements to a page and measuring 0.62% engagement is the Italian study's finding, and it is what most hospital Facebook activity amounts to.

Use it if: you serve an older demographic, manage chronic conditions, or run a community hospital where local presence matters. Do not use it as your primary channel for a subspecialty practice whose patients travel.

Instagram — visual specialties, and the biggest regulatory trap

Audience: patients under 45. Aesthetic, dermatological, dental, orthopaedic, physiotherapy, fertility, obstetrics.

Survey data from Hall & Partners and Think Next found that 33% of Gen Z patients use social groups to discuss an illness, naming TikTok, Instagram and X as their top three platforms. Widely-circulated figures claiming a majority of Instagram users search it for health information trace to content-marketing aggregators and we have excluded them.

The genuine job it does: demonstrating things that are visible. A physiotherapist showing a movement pattern. A dermatologist explaining what a lesion looks like. Anything where the clinical reality is legible in an image.

Where it fails, and this is the important part: Instagram's native grammar is before-and-after, transformation, and testimonial. In Australia that grammar is largely prohibited — Section 133 bans testimonials referencing clinical outcomes, before-and-after imagery is heavily restricted, and new cosmetic advertising rules effective 2 September 2025 added a ban on influencer testimonials and on advertising cosmetic procedures to under-18s. In India, the operative code prohibits soliciting patients directly or indirectly.

So the platform most suited to visual specialties is the one whose default content format is the one those specialties are least permitted to use.

The compliant version is better anyway: process, eligibility, technique, and who you would not treat. It converts worse per post and it is the only version that survives a regulator and gets cited by an answer engine.

LinkedIn — the referrer and the institution

Audience: referring doctors, hospital administrators, health system executives, recruitment. Almost no patients.

This is the platform most consistently underused by specialist practices, and the one whose audience matters most to them.

Physician survey data indicates 35% of doctors find LinkedIn the most helpful platform for following key opinion leader presentations and policy developments. For B2B healthcare — reaching administrators, executives, and referral partners — it is the only serious option.

The genuine job it does: making a specialist legible to other doctors. Case selection criteria. What you take and what you decline. What a referrer receives back and when. These are the four things a referring GP actually wants to know, and almost no practice publishes them anywhere.

Where it fails: as a patient channel. Patients are not on LinkedIn looking for an oncologist.

Use it if: you are a specialist practice dependent on referral — which is most of them. This should probably be your first platform, and it almost never is.

X — the profession's public square, and a shrinking one

Audience: clinicians, academics, journalists, conference discourse.

Physician data shows 62% use X to follow hospitals and professional organisations.

The genuine job it does: professional visibility among peers. Conference commentary, paper discussion, society engagement. It is where a consultant becomes known to other consultants, and being known to other consultants is where referrals originate.

Where it fails: patient acquisition, almost entirely. And the platform's volatility since 2022 has fragmented the medical community — a meaningful share has migrated or gone quiet.

Honest assessment: high value for a narrow purpose, declining reliability. Worth maintaining if you already have standing there. Not worth building from zero in 2026 unless your specialty's academic conversation demonstrably still lives there. Check before investing.

YouTube — the most underrated, by a distance

Audience: everyone. Patients, referrers, and peers all use it, at different moments.

YouTube functions simultaneously as a social platform and the second-largest search engine, which makes it the only channel serving all three audiences with the same asset.

Why it is the strongest fit with the signalling argument: video is the format in which a clinician's expertise is least fakeable. Two minutes of a surgeon explaining case selection carries information no written copy conveys — hesitation, precision, the way they qualify a claim. It is expensive to produce in the sense that matters: it requires someone who actually knows the answer.

It also compounds. A surgeon explaining their approach in 2026 is still explaining it in 2030. Nothing on Instagram survives a fortnight.

And it is indexable, with transcripts feeding both classical search and answer engines.

Production cost is lower than assumed: a quiet room, a phone on a tripod at eye height, a lapel microphone, a window. Audio matters far more than image quality. An afternoon yields six to eight clips if the questions are prepared.

If a practice can only do one thing, we would argue for this one.

Google — not social, and the largest single input

Not a social platform, but it belongs in any honest media comparison because it is where the decision usually resolves.

Search is where most clinical decisions get checked. We have deliberately not quoted a headline percentage here: the figures in circulation for "patients who research providers online" trace back through several layers of citation to studies we could not locate, and the number does not change the argument.

Two distinct jobs:

Google Business Profile decides proximity queries. If your patients search "dermatologist near me," the map pack decides the outcome and no amount of Instagram changes it.

Search and AI Overviews decide condition queries. The patient who has been diagnosed and is researching does not search your name. They search the diagnosis, the treatment, the drug as spelled in their report. This is where entity work, authored content and credential markup operate.

The critical distinction: work out which of your queries are proximity-decided and which are content-decided before allocating anything. They require opposite work and most practices need both, weighted very differently.

Walled gardens — Doximity, Sermo, and their equivalents

Frequently omitted from media plans and, for peer reputation, among the most consequential.

Physician survey data puts Doximity at 59% and Sermo at 58% for clinical and professional purposes — above every public platform.

These are peer channels, not patient channels, and they matter because peer reputation is upstream of referral. In India the equivalent is not a platform at all but WhatsApp groups — specialty associations, alumni networks, hospital departments. Treating WhatsApp as informal is a category error; in Indian medicine it is where professional conversation actually happens.

The comparison, condensed

Combinations, by practice type

Media selection is a stack, not a channel. Four that we would actually recommend.

Subspecialty consultant whose patients travel

Oncology, nuclear medicine, complex cardiac, rare disease

Core: YouTube + LinkedIn + Google Search. Patients arrive diagnosed and researching, or by referral. YouTube carries the expensive signal and is indexable. LinkedIn reaches the referrers who supply most of the volume. Search captures condition-level intent. Skip: Instagram, Google Business Profile as a priority. Your patients are not choosing on proximity.

Local general practice or community clinic

Core: Google Business Profile + Facebook. Proximity decides these queries. GBP completeness, category accuracy, NAP consistency and a genuine review practice will outperform everything else combined. Skip: LinkedIn, X. Add Instagram only if you have a visual service line.

Visual specialty

Dermatology, aesthetics, dental, orthopaedics, physiotherapy, fertility

Core: Instagram + YouTube + Google Business Profile. The only stack where Instagram earns first position — but built on process and eligibility content, not transformation, because the transformation format is the one you are least permitted to use. Watch: this stack carries the highest regulatory exposure of any. Audit before you publish.

Hospital or multi-specialty group

Core: YouTube + Facebook + LinkedIn + Google, run as four distinct channels. The Italian study's 82.3% cross-posting finding is the exact failure to avoid. LinkedIn for recruitment and B2B — which for a hospital is a genuine and separate business need. Facebook for community. YouTube for the clinicians. Google for intent. And resource it properly or do less. Four channels done identically is worse than two done distinctly.

Five conclusions, and one open question

1. Choose the audience before the platform. Patient, referrer, or peer. Almost every plan we review chooses platforms first, then discovers it has been talking to nobody in particular. If you are a specialist, your referrer audience is probably worth more than your patient audience, and it is the one you are almost certainly not addressing.

2. Match format cost to credential. The signalling logic is unambiguous. A consultant's authority derives from an expensive, hard-to-fake credential. Cheap formats undercut it. This does not mean high production values — it means visible intellectual effort. A phone-shot twenty-minute explanation is expensive in the way that counts. A polished fifteen-second Reel is not.

3. Regulation removes the easiest tactics, and they were the weak ones anyway. Testimonials, before-and-afters and superlatives are prohibited or restricted in both markets. They are also not citable by an answer engine, carry no extractable claim, and are indistinguishable from every competitor's. The compliant substitute — the clinician explaining case selection — is better on every dimension except ease.

4. YouTube is systematically underweighted. It is the only channel serving patients, referrers and peers with one asset, the only one where the asset compounds for years, the only one indexable by both search and answer engines, and the one whose format best matches the signalling requirement.

5. Do fewer, distinctly. 82.3% identical cross-posting at 0.62% engagement is what breadth without resourcing produces. Two platforms with genuine platform-native content beat five with the same content pasted five times.

And the open question we cannot yet answer. Almost all the platform data available is from 2022–2024 and predates AI-mediated search at scale. If patients increasingly ask a model rather than search or scroll, the value of social presence may shift from reach toward corroboration — social profiles functioning primarily as sameAs evidence that a clinician is who they claim to be, rather than as an audience channel in their own right.

We think that is where this goes. We do not have the data to prove it, and anyone telling you they do is ahead of the evidence.

Sources

• Fiammenghi C, Covolo L, Vanoncini A, Gelatti U, Ceretti E. Institutional health communication and social media: Exploring Italian hospitals' use of social media pages. Digital Health, 2024. N=194 public hospitals; 36.6% with a profile, 18.3% with a policy, 82.3% cross-posting Instagram/Facebook unmodified, 0.62% mean engagement.
• Sermo / LiveWorld physician survey, November 2022 (200+ US physicians across specialties) — Doximity 59%, Sermo 58%, Facebook private groups 52%, X for organisations 62%, LinkedIn for KOL and policy 35%, 57% changing perception of a treatment based on social content.
• Hall & Partners / Think Next, 2022 — 10,500 adults across China, Germany, Japan, UK, US. Chronic-condition patients citing doctors as primary source 62%; 38% turning to social and influencers; Gen Z top three platforms TikTok, Instagram, X.
• Healthcare Success, platform guidance 2026 — Facebook and Instagram for patient-facing acquisition; YouTube as platform and search engine; LinkedIn for HCP and B2B. (Vendor source; used for directional platform-role framing only, not for any statistic.)
• Sutherland R. Alchemy: The Dark Art and Curious Science of Creating Magic in Brands, Business, and Life — psycho-logic, costly signalling, satisficing, psychophysics; the Knowledge as trust placebo; friction as credibility.
• AHPRA advertising guidelines, including cosmetic advertising rules effective 2 September 2025.
• Indian Medical Council (Professional Conduct, Etiquette and Ethics) Regulations, 2002 — operative following the abeyance of the NMC 2023 regulations.

A note on data quality, and what we removed. Most healthcare social media statistics circulating online come from content-marketing aggregators citing each other. Figures contradict within single sources — we found one page giving both 89% and 95% for the same metric on the same screen.

An earlier draft of this article used six figures from those sources: US hospital Facebook adoption, Instagram health-search share, daily health posts on X, Doximity physician penetration, and two provider-selection percentages. We cut all six, because an article criticising aggregator data cannot then rely on it. What remains comes from one peer-reviewed study and two named, dated surveys with stated sample sizes.

The result is fewer numbers. We think it is a better article for it, and if a claim below matters to a decision you are making, the citation is there so you can check it rather than trust us.

Several of the platform surveys date from 2022–2023 and should be treated as directional rather than current.

Pontem One works exclusively on clinical accounts in India and Australia.