Healthcare Is A Vertical. Health And Its Data Are Horizontal

Healthcare is a vertical. Health and its data are horizontal.

Why your next healthtech pilot will stall, and why it's not your technology's fault

Yesterday I had the pleasure of speaking at HealthTech Adria 2026 in Split. A room full of builders: founders, tech-transfer people, innovation leads, hospital leaders and investors from across Southeast Europe. Marko Mimica opened the day with a sharp thesis, No Adoption, No Innovation. Agathe Acchiardo took us to 2035 and reminded us that none of it needs a new invention. It needs a system that can absorb it.

My task was to make that actionable. So I started with a question I've asked boardrooms and ministries on several continents.

What percentage of our health is actually determined by healthcare?

The room guessed high. They usually do. The answer, from the NEHI analysis I've been showing for years: access to healthcare accounts for roughly 6% of what drives our health. DNA about 20%. Postcode, where you live and work, about 22%. Behaviour 37%. And yet around 90% of what we spend goes to medical care.

Let that sink in for a second, because it changes who you should be talking to.

Health is mostly created outside the walls we build our products for. And that is exactly why so many good healthtech solutions never leave the pilot stage.

Verticals within verticals

Healthcare is organised in verticals. The hospital. The GP. The pharmacy. Home care. Mental health. Public health. The payers.

And inside each vertical, we built another layer of verticals. Departments. Specialties. Disease-oriented groups. Budgets per unit. Each with its own tool, its own data and, yes, its own pilot.

That worked, more or less, as long as information stayed where it was created. That era is over.

We are entering an era of ubiquitous data. Data from the wearable, the pharmacy, the lab, the insurer, the smartphone, the employer's health programme and the patient's own AI assistant. That data doesn't respect our walls. It seeps through every vertical, and through every vertical inside it.

Agathe's patient story made it tangible: Cathy, 49, goes from a routine cholesterol result to an osteoporosis pathway in one hour, through an app, a pharmacy hub, AI, a nurse practitioner and her GP. Five walls crossed in sixty minutes. Today, each of those walls runs its own pilot.

AI is a horizontal technology. We keep buying it vertically. That's the paradox behind the number I opened my essay with: 78% of healthcare organisations use AI (Deloitte, 2026), yet nearly two-thirds haven't scaled it across the enterprise.

The next winner in healthcare isn't a vertical. It's the layer underneath: data, identity, compute, governance and trust. And right now, nobody owns it.

Welcome to pilot purgatory

So why do we keep piloting? Because a pilot is comfortable. I showed the room my list, and I saw a lot of knowing smiles:

• No difficult decisions, and separate departmental decisions (or simply a new pilot).
• Not owned by the business. Compliance not involved from the get-go.
• A performance case instead of an impact case. No value case. No K2Is, Key Impact Indicators.
• No (regional) collaboration: the not-invented-here syndrome.
• No public "disgrace", because it was only a pilot. And no effect on anyone's own role.
• And the question almost nobody asks: what will we stop, if it works?

Then there's the 8-ounce cup problem. The cup is already full. Clinicians, nurses and administrators have no room left. Pouring AI on top doesn't add capacity; it spills. Solve the actual problem first, and make room before you pour.

A pilot is a safety net; in a pilot you're quietly looking for reasons to stop. A plot is a roadmap. One tests the water, the other builds the bridge. That's why I keep saying "implementation test" instead of "pilot". It's more than a word game.

But here is the part I want to underline today. Many pilots don't stall because the technology fails. They stall because the people who decide whether it survives were never in the room. Marko's own numbers for Croatia made the point painfully clear: among the organisations in 210 funded health-innovation projects, 92 were startups and companies, 23 research institutions, and only 16 hospitals and clinical partners. The adopters were the smallest group at the table. And I'd argue the table was still too small.

Change comes from the edge

Here's my provocation to every healthtech builder: if you only talk to healthcare providers, your lens is too small.

Change in healthcare has rarely started at the centre. It comes from the edge, and increasingly from outside-in. Retail moved into health with clinics, pharmacies and consumer-grade experiences. Big tech built health into the operating system on your phone. Insurers and employers fund prevention programmes. Municipalities run social care and the places where health is actually created. Patients walk in with their own data and, more and more, with their own AI.

The structure is shifting too: from centralised, to decentralised, to distributed. From what I call an eGo-system, organised around institutions and their data silos, to an eCo-system organised around the life of the person.

Now look at a typical stakeholder map for a healthtech pilot. A department head. A CMIO. Maybe procurement, late in the process. Where are the pharmacist, the insurer, the municipality, the employer, the patient organisation, the consumer platform the patient already uses every day? Where's the finance ministry, for that matter? Bloom, Canning and Sevilla showed that one extra year of life expectancy can lift economic growth by around 4%. The Lancet Commission on Investing in Health put the potential return on health investment as high as nine times. Health is an economic agenda, not just a care agenda.

When those stakeholders aren't at the table, the pilot can be clinically brilliant and still die. Nobody outside the vertical owns its continuation. Nobody pays for the layer underneath. Nobody connects it to the place where the patient actually lives.

That's why I told the room I was slightly worried I was preaching to the wrong choir. Most builders already want adoption. The bottleneck isn't in your code. It's in the system you sell into: boards, funders, ministries, procurement, payers, EU programmes. But builders can do one thing the system rarely does on its own. You can widen the table.

What to carry out of the room

I left the builders in Split with three questions for every buyer, funder and ministry they meet. Today I'm adding a fourth, the outside-in one:

1. Which horizontal layer will this run on, and who owns it?
2. Who is accountable for the outcome, and when do we decide to deploy or stop?
3. Which health outcome will you report to your board?
4. Who outside healthcare is already solving this problem, or already holds the data, the channel or the trust to make it stick?

If nobody can answer the first one, you've found the real project. It just isn't yours to build alone. And if nobody has even thought about the fourth, you've found the reason your pilot will stall.

For the system leaders, the moves are bigger: design the operating system instead of buying the next app, make every innovation accountable for its outcome, treat the foundation as a board-level capital decision rather than a project grant, and measure health instead of activity. I wrote about that in my essay AI Is Not a Product. It's Infrastructure. and in my newsletter piece AI Infrastructure is not a technology decision; it's a clinical decision.

Marko asked the room what it would take for innovation to be adopted. My answer: an owned horizontal layer. Designed as a system. Accountable for outcomes. Governed by the board. And built for health, not only for healthcare. Special shout out to Aline Noizet for clarifying crips and clear why procurement needs to sit at the table as well And my friend Tony Young teasing the panel in a way only a Brit can do ;-)

Because in the end, trust was always the real product. AI, built as infrastructure, may finally be the way to deliver it at scale.

Not because it's innovative. Because it's existential.

Thank you, HealthTech Adria, NetHub, Marko, Agathe and Jordi Serrano Pons and everyone in the room in Split. If you want this conversation in your boardroom or at your event, you know where to find me.

So, over to you: who is missing from your stakeholder map? Tell me in the comments.

\#HealthTechAdria \#DigitalHealth \#AIinHealthcare \#PilotPurgatory \#HealthcareTransformation \#Innovation