A small fund with a healthcare allocation faces a structural problem that has nothing to do with talent. The partners are good investors. The associate is capable. Nobody on the team has read a 510(k) submission, sat on a data safety monitoring board, or run a design history file — and the deals require someone who has.
There are three ways funds cover this. Each works for a real set of questions and fails for a different one, and the failure modes are not symmetric. Worth being precise about, because the usual framing — “do we have the expertise or not” — is the wrong axis.
Option one: the generalist associate
The associate reads the deck, works the public databases, builds the model, and writes the memo. This covers more than people assume. Clearance status, corporate history, patent posture, registry entries, comparable transactions — all of it is public, and a careful associate with a method gets most of it right.
Where it reaches its limit is not effort. It is that the questions which decide biomedical outcomes are questions you have to already know to ask. Whether a chosen predicate is defensible for this indication. Whether a proposed endpoint will satisfy a payer as well as a reviewer. Whether a design change since clearance was significant enough to warrant a new submission. Whether the failure mode a device addresses is the one its architecture actually engineers against. None of these is a lookup. Each is a judgment built on having done the thing before, and an associate cannot be faulted for not making a call they have no basis to make.
The honest description of this lane: excellent coverage of what is documented, structurally blind to what is not asked.
Option two: the expert call
An expert network puts a clinician or a former regulatory lead on the phone for an hour. The depth is real and immediate — this is the fastest way to get a genuinely informed read on one question.
Three things constrain it, and they compound. The first is scope: an hour buys one perspective on the questions you knew to ask, which returns you to the previous problem. The second is documentation — a call produces notes, not a citable record, and notes do not travel well to an IC, a co-investor, or an LP asking a year later what was known at the time. The third is the quiet one: the specialist you reach is deep in a specialty, and the deal spans clinical, regulatory, reimbursement, engineering, quality and IP. Coverage across those domains is not what a single call is shaped to give.
It is a sharp instrument aimed at one point. That is genuinely useful and it is not a diligence process.
Option three: a specialist review
A structured outside assessment covers the domains systematically and produces a document. What varies enormously — and what is worth interrogating before buying any of it — is whether the document is checkable.
A memo that asserts a regulatory pathway is a judgment you are asked to trust. A memo that asserts a regulatory pathway and cites the K-number, the product code and the indication as cleared is a judgment you can verify in four minutes. Those are different products, and the difference only becomes visible when something turns out to be wrong.
The second question worth asking is what the assessment compares against. A read on a single company in isolation is worth less than the same read positioned against companies that faced the same pathway, because almost every useful question in this space is comparative: is this timeline normal, is this evidence package thin, is this reimbursement route the one others took. That requires an assembled base of comparable companies, which is the part that cannot be produced on demand for one engagement.
What none of the three can do
It is worth being equally clear about the ceiling, because the industry is not.
No process — ours included — tells you which company will succeed. Early-stage biomedical outcomes turn on trial results that have not happened, reviewer judgments not yet made, and competitive and financing conditions nobody observes in advance. Anyone selling a probability of success is selling a number whose derivation will not survive contact with a statistician. We tested this on our own substrate and it is the reason we do not make the claim.
What diligence can do is narrower and genuinely valuable: establish what is documented, establish what is not, put both in front of a decision-maker with sources attached, and surface the specific things a sophisticated reader would have wanted to know and would not have found on their own. The decision stays where it belongs.
The composite most funds land on
In practice the three are not alternatives. The associate runs the public record and the model. The structured review covers the domains and produces the citable base document, including the questions the team did not know to ask. The expert call gets aimed — usefully, now — at the two or three specific uncertainties the review surfaced rather than at a blank page.
That ordering matters more than the components. An expert hour spent on a question you already knew to ask is the most expensive way to confirm something. The same hour spent on a question a systematic review just put in front of you is the cheapest resolution of a real unknown you will buy all quarter.
Where Vantage sits
We are the third lane, built by a physician and a biomedical engineer, and deliberately narrow about what we claim. Every factual assertion in a Vantage assessment is cited to a primary source — FDA, SEC, ClinicalTrials.gov, USPTO, CMS — and each target is positioned against comparable companies drawn from a substrate we have assembled rather than assertions drawn from memory. Where the record is thin, the report says the record is thin and names what would resolve it.
Funds can run a single deal first or license the platform annually for a set number of Quick Screens and Diligence Reports. A license suits steady deal flow, because the cost that matters most to a small team is not the price of a report. It is the partner-hours consumed by deals that do not clear the first screen. Details of the license tiers are on the investors page, and a sample deliverable will tell you more about whether it is checkable than any description of it will.
References
- FDA. “Search Databases” — index of FDA public databases used in device and drug diligence. fda.gov
- FDA. “510(k) Premarket Notification” database. accessdata.fda.gov
- U.S. Securities and Exchange Commission. “EDGAR Full Text Search.” sec.gov
- ClinicalTrials.gov. “FDAAA 801 and the Final Rule.” clinicaltrials.gov
- Centers for Medicare & Medicaid Services. “Medicare Coverage Database.” cms.gov
See Whether It Is Actually Checkable
Open a sample deliverable and try to verify a claim in it. That is the test worth running on any diligence product, including ours.