Traditionally, the one skill differentiating a great vs mediocre healthtech company was sales & marketing. It’s the first lesson many founders learn when they enter the field. It’s much more about a reliable sales funnel than evidence or a great product. Doctolib, Docplanner, Oura… all multibillion companies that are ultimately sales and marketing machines.
Now, in times where AI products rule the market, that’s changing! Sure, sales remains an important topic, especially with all the competition around.
But now that many hospitals, clinics, and other slow institutions finally switch to new software, another topic becomes a bottleneck: implementation and rollouts. How do you get all these shiny new AI tools to actually be used? At a reasonable time scale? Without frustrating people? How do you convince Wolfgang (57) from IT that, in fact, your AI is not a secret backdoor for Peter Thiel?
Today, we’ll explore exactly that: The why tech implementation matters more than ever for the startup scene, and ideas how to navigate it. Useful obversations from the field.
Note: I’m speaking from a European perspective throughout this article - although in the US, the blockers are likely similar. Startup funding and rollout timelines might differ though. The EHR landscape is another important difference, where Europe has a more fragmented market and EPIC as a vendor is less dominant. Just keep that in mind.
AI implementation will determine the winners
I’ve been sitting in many events, calls and coffee chats recently where people complain about hospital implementation. Interestingly, the more successful the startup, the more relevant the topic.
It’s not the slow and struggling companies, but rather the hot hospital AI darlings (you know who) that face implementation challenges. You can essentially be certain that all health AI companies that have raised €50m+ struggle with getting their products used consistently across their customer bases. This is not a “loser” debate!
Most of these companies have abundant funding, and dozens of initial contracts. They’ve raised their massive Series As on the promise to “land and expand” from here on out. Hence, the valuations don’t reflect their current revenue, because that’s often a lofty 20-30x multiple. Fulfilling these valuations with newly closed contracts is tough. Instead, the valuations often rest on the rollout potential across hospitals, sites, or departments. It comes down to: Will they nail implementation or not?
That’s also reflected in how they hire - Tandem Health & Co have a significant pool of people that babysit hospital onboarding. And since “hospital baysitter” is not a cool job title, they usually call them “Medical Operations”, “Medical Consultants”, etc., fair enough.
Five challenges, and ideas how to solve them
Throughout all my discussions with founders, I’ve collected a bunch of typical problems in hospital adoption, along with do’s and dont’s. Think of this as a checklist - it’s also what I grill founders on in due diligence calls.
To be clear: I made none of these lessons up myself, all of them are rooted in stories that operators or hospitals told me. I’m just a humble obvserver collecting them. They all stem from hospitals and sometimes smaller clinics across Europe.
Lesson 1: Chaotic procurement & forgotten stakeholders
Hospitals sometimes have no clue how to buy your product. That applies especially to newer technologies that are not standard medical equipment or supplies. Your internal champion might run around like a headless chicken trying to get sign-off.
During implementation, that can lead to “forgotten” stakeholders! One company I know found out only after contract signature that they forgot to involve an entire department. These guys were completeley confused by the product appearing on their desks, and they had to park rollout for months.
Solution? Sounds strange, but founders need to basically teach their customers how to buy and implement their AI products. It’s worth checking diligently for every little stakeholder you need onboard. Do you need to involve logistics or sterilisation departments for example? Do that sooner rather than later.
Lesson 2: Digitisation backlogs
You would not believe the amount of digitisation debt in European clinics. I can definitely speak for Germany! Sometimes pen and paper is still the standard, sometimes no internet connection in parts of the building. That can bring unexpected blockers to an implementation process. As a side effect, customer IT capacity might be clogged with quite basic transformation projects. One company I know recently closed a deal with a German healthcare institution, only to find out that employees had no email addresses - unfortunate if that’s your standard authentication path...
Solution? Some founders tell me they stay extremely close to their customer’s IT. Build a relationship, get to know their families, support them where they can. Others told me they’ve seen their IT counterparts go into burnout after the implementation… Pro tip: You wanna avoid that.
Lesson 3: Data migration nightmares
The longer a clinic is around, the worse of a challenge is data migration. This applies especially to any software aiming to become a main data repository (EHRs, practice management systems, …). A major German outpatient clinic (12+ physicians) recently told me: “At this point we can’t, under any circumstances, switch our PMS/EHR. The data migration has become impossible at our scale”.
Solution? Some startups offer dedicated migration services, especially for very common incumbent systems. LLMs have arguably made migrations easier overall. But I don’t see an easy solution here - let’s just say the inertia shouldn’t be underestimated.
Lesson 4: Pilots, pilots, pilots
Hospitals love to internally test or pilot products. They rarely trust external benchmarks. One large hospital chain in Berlin for example has a policy that every external clinical product gets its own internal benchmark. That creates delays. It’s generally okay - as long as it actually results in a paid contract.
The worst kind of projects are unpaid ones with an academic angle. Researchers love to abuse startups for projects that result in academic publications for them, but never in a real contract. If your hospital customer seems to prioritize publishing papers over real-life commercial ROI, run.
Solution? Step one: Plan for it, and be honest to yourself! The best founders are truffle pigs that sniff out attractive pilots vs useless ones. Sometimes there’s no way around it. Step two: Ideally, you only take paid pilots and straightout ignore everything else. As an investor, I discount unpaid pilots completely for that reason.
Lesson 5: Internal (academic) competition
Many customers won’t tell you, but especially in academic hospitals there’s often an internal research team working on something similar to your product. AI papers get great impact factors, so many academic hospitals have AI research teams. Even if they’re never pushing their products into production, you’re stepping into their territory and they might see you as competition.
Solution? Be very thorough in your prep. Find those teams and appease them before you sell. This can require some Machiavelli-style thinking.
Lesson 6: Timelines?
One more interesting learning in general: A top operator in the field told me about his approach towards implementation timelines. Their company usually frontloads implementation work - starting before contract signature.
So while many startups treat it as a sequential process (first secure the sale, then start implementation), it might be worth kicking off awareness campagains and other change management processes in the final sales stages already. To compress the overall timeline. You risk investing your own resources prematurely, but it might give you an edge.
Is this a business opportunity?
I can’t help but wonder whether this isn’t an opportunity in itself. Most startups will (rightfully) try and build internal armies to steer rollouts, but for large transformations you’ll need external support.
Ironically, the most profitable players in the game might not be the AI companies, but the implementation consultants.
It’s btw great news for society: No matter how many jobs are displaced by AI, I’m confident they will all find work in hospital implementation…
Speak soon,
Lucas
P.S.: I’m curious to hear of any further tips, and from any consultants riding that wave!
Lesson 1 and Lesson 4 sit oddly together. The company that found a forgotten department after signature had money already committed and still no settled rollout decision.
Which is why paid versus unpaid may be doing less work in diligence than it looks. Payment shows a budget exists somewhere in the building. It does not show that anyone with authority over the rollout agreed in advance what result would move them. An innovation budget answers that no better than a research grant.
So maybe ask who agreed, before the pilot started, to act on the outcome.
The implementation consultants may be the real business opportunity here. AI companies can build a great product and sign the hospital, but a contract is not the same as adoption. Someone still has to align departments, navigate IT and procurement, fit the product into existing workflows, train people, and make sure the pilot becomes something the hospital actually uses. That may be where a lot of the value ends up sitting. Healthcare loves innovation as long as it requires no new login, no extra click, and absolutely no change to anyone’s workflow. Implementation consultants should have excellent job security.