Health Tech Founders, You’re Selling to the Wrong Room: the Principal Agent Problem.

The principal-agent problem is the central unsolved equation of health tech go-to-market — and the EHR makes it exponentially harder. Here’s what the structure looks like, and what you can do

Building Techy Surgeon’s audience taught me something unexpected about healthcare distribution. When I write for surgeons, I get surgeons. I write AI tutorials and I get a mix of anyone interested in AI. When I write for VCs, I get investor types. The feedback loop: write something true, and the people who live that truth find you.

Selling to health systems is the opposite of this. You can build something a surgeon desperately needs, demonstrate it to the surgeon who desperately needs it, get that surgeon to write an internal memo about how desperately they need it. And you will still lose the deal. Not because the product was wrong. Because you were selling to the wrong room.

I’ve sat in the room where this happens. The surgeon champion, the person whose workflow would actually change, sits at the far end of the table. The buying committee sits at the near end. The champion speaks maybe once. The conversation is dominated by people who will never use the product asking questions about things the product was never designed to address: total cost of ownership, IT integration risk, HIPAA audit trails, EHR compatibility. You leave the meeting thinking it went well. Three months later, a slow, polite email arrives. The initiative has been “deprioritized.”

This is not bad luck. This is a structural feature of health system procurement with a name economists have studied since Jensen and Meckling wrote about it in 1976 in their foundational work on the theory of the firm. It is called the principal-agent problem.

The Problem Has a Name

In the formal definition: the principal-agent problem occurs when an agent makes decisions on behalf of a principal, but the agent’s incentives, information, and priorities do not perfectly align with the principal’s. The classic examples are shareholders versus management, or clients versus lawyers. In healthcare procurement, the structure is more layered and more treacherous.

Brendan Keeler — who writes the essential Health API Guy Substack and understands the infrastructure of health data better than almost anyone — framed this precisely when he described the problem as “priorities conflict between the buyer/owner of an asset and the one to whom the owner delegated control.” In his piece “There will be Bundling” he walks through how product adoption in health systems gets stuck not in the hands of the people who would benefit from the product, but in the hands of those to whom purchase authority has been delegated. Those delegates have their own priorities, their own risk calculus, their own institutional pressures — and they rarely map neatly onto the clinical problem you set out to solve.

“You can build something a surgeon desperately needs, demonstrate it to the surgeon who desperately needs it — and still lose the deal. Not because the product was wrong. Because you were selling to the wrong room.”

Keeler’s An Epic Tale: The Startup Odyssey is also worth your time. In it, he maps the specific geography of the Epic integration odyssey and how the structural gatekeeping compounds the buyer/user misalignment in ways that are unique to health systems. More on that below.

Three Floors. Three Conversations. One Building.

Most health system sales fail because founders treat the building as one room. It is three floors. The conversations you need to have on each floor are different; the language is different; the evidence that moves the room is different. Confusing floors is where deals go to die.

The first floor is clinical. This is the surgeon, the nurse practitioner, the care coordinator, the physical therapist — the end user whose daily workflow the product actually changes. These people can tell you immediately whether your product solves a real problem, and they will often become your fiercest internal advocates. But they almost never control budget. In the traditional B2B procurement motion, winning the first floor is necessary but not sufficient. The trap is assuming that a passionate clinician champion will carry the deal upward. They rarely do — not because they don’t try, but because the system is not designed to route clinical enthusiasm into purchase authority.

The second floor is clinical and operational leadership. This is the CMO, CNO, department chair, VP of Quality. These people care about outcomes — patient satisfaction scores, readmission rates, throughput, workforce burnout. They are more likely to have discretionary budget and more likely to be measured against the metrics your product might actually move. This floor is where your outcomes story needs to be crisp. The a16z enterprise buyer analysis puts it clearly: clinical leaders want to see hair-on-fire problems addressed with demonstrable, measurable outcomes. Abstractions do not work here.

The third floor is where deals are approved — and killed. IT security, procurement, legal, compliance, the CIO’s office. This floor is not interested in your clinical outcomes story. They are interested in risk. SOC 2 Type 2 certification, HIPAA Business Associate Agreements, EHR integration scope, data residency, liability language. If you arrive here without having done this homework, the deal does not stall — it ends. And it ends with people on the third floor looking reasonable, because their job is literally risk management. Brian Manning’s piece Why Is It So Hard To Sell To Health Systems documents this dynamic with painful clarity: procurement timelines in health systems routinely run 6–18 months, and the third floor can extend them indefinitely.

And Then There’s the EHR

If the principal-agent problem is a multi-headed beast, the EHR is the environment that beast lives in — one that actively feeds it. I’ve written about this before in “Yes the EHR is Going to Kill your Startup”

EHR integration is not an implementation detail. It is a strategic decision that determines whether your product gets used, gets tolerated, or gets quietly removed.

The integration hierarchy matters enormously here, and founders too often discover it too late. Athenahealth and Modmed offer relatively open APIs with manageable complexity. Cerner has improved substantially with Oracle’s stewardship. The long tail of smaller EHRs varies wildly. And then there is Epic — the platform that commands the majority of large academic medical center market share — which, as Brendan Keeler documents in extraordinary detail, has built a procurement and integration architecture that compounds the principal-agent problem structurally.

The specific mechanism Keeler describes for Epic Community Connect hospitals is worth understanding: in these systems, the parent health system’s IT organization becomes a gatekeeper for the affiliated hospital’s procurement. This means the agent relationship adds a third layer; The end user, the local administrator, and now the parent IT organization, each have their own priorities, timelines, and risk tolerance. The product has to pass all three filters. The integration requirements that were not scoped pre-sale become the landmine that ends deals that should have closed.

The Overlooked Frontier

Buried inside all the complexity of health system sales is an underappreciated opportunity. The large AMC is, I’ll say it plainly, a laborious slog. Success there requires serendipity, political capital, an internal champion with sustained tenure, and often an inside connection who knows which floors to skip. The timeline is long. The implementation risk is high. The political complexity is substantial.

But the community hospital under 250 beds, the FQHC, the rural health system…these are a different story entirely. The principal-agent distance is shorter. The CMO may also be the clinical champion. The procurement process has fewer floors and thinner walls between them. The EHR environment is often more flexible. And the need is, in many cases, more acute: these organizations carry disproportionate patient complexity and have fewer internal resources to address it.

The innovation frontier in healthcare is not the prestige AMC deployment that makes a compelling press release. It is the 180-bed community hospital that desperately needs what you built and actually has the organizational agility to use it.

The Strategic Move

The full picture is this: health system sales is the hardest go-to-market in existence when the principal-agent problem is fully loaded. You’re navigating a three-floor decision architecture, an EHR integration environment that may add additional gatekeeper layers, procurement timelines that can absorb a startup’s entire runway, and a buyer who is not the user. The a16z analysis frames this well — the enterprise healthcare buyer is fundamentally risk-averse in a way that commercial buyers in other sectors are not, because the cost of a bad implementation is not just financial but reputational and regulatory.

This does not make health system sales impossible. It means the founders who succeed there have mapped the building in advance, have evidence for each floor, have done the compliance homework before it becomes an emergency, and have developed a realistic picture of which building to try to enter first. The B2B-only motion is viable in health systems with the right sequencing. The B2B2C motion — like what we’re navigating with the CMS ACCESS Model — adds patient-facing dynamics that create value at the first floor while making the case at the second. And increasingly, the answer for early-stage companies may be to bypass the building entirely: to become a care model, a Part B organization, a clinical entity that demonstrates outcomes through its own delivery before asking a health system to absorb the risk.

What remains unsettled is whether value-based care will structurally repair this misalignment over time. In a world where health systems are genuinely held accountable for total cost of care and outcomes, the principal-agent problem shifts: the third floor has a reason to listen to the first floor, because the first floor’s clinical decisions now have financial consequences at the organizational level. That world is arriving — slowly, unevenly, and with enormous regional variation. The operators who build for that world now will be positioned for when it lands.

What follows is for paid subscribers: a six-point operator playbook for navigating health system sales with the principal-agent problem in view, built from direct experience across B2B, B2B2C, and care model deployments.

⭐ Subscriber Playbook

The Six Moves I Believe Founders have at their Disposal