Market access is an intelligence problem.
Market access is an intelligence problem
A health technology company wants to sell into a new market. Before anyone writes a proposal, five questions need answers.
Who actually buys this, and through which route? Which framework or tender sits behind that purchase? Is the product regulated, and what is its current status? Who else is already in the market, and what do they hold? Is there clinical or research activity that changes the picture?
Each question has an answer. Each answer lives in a different place. A procurement notice sits in one system, a company record in another, a regulator's register in a third, a trials registry in a fourth. They update at different speeds, they describe the same organisations under different names, and none of them tells you how old its answer is or how far to trust it.
The company does what companies always do. It assembles the picture by hand, from tabs and spreadsheets and a few expensive conversations, and it hopes nothing changed while it was assembling.
The tools are good. The gaps are between them.
There is no shortage of intelligence tools in health. Procurement portals, company registers, regulator databases, clinical registries, market reports. Many of them are very good at the one thing they do.
The trouble starts at the joins. A supplier appears in a tender notice under a trading name, in a company register under a legal name, and in a regulator's database under a third variation. Whether those are one organisation or three is a question every user answers separately, privately, and differently. A regulatory status is correct on the day it was read and wrong a fortnight later. A ranking looks like analysis and may be partly something else.
For a large company with a market access team, this is an irritation. For a startup with six people and one chance at a first contract, it can decide where the effort goes and whether it is spent in the right place.
None of these are failures of the individual tools. They are what happens when good tools are used without a layer that holds the context between them.
What the layer has to do
Bringing the best tools and sources together is not an aggregation exercise. Aggregation puts everything on one page and leaves the disagreements for the reader to sort out. The layer has to do four more specific things.
It has to resolve entities. A supplier, an NHS body and a regulated product have to be recognised as the same thing wherever they appear, with the resolution logic recorded. That is the job Makemake exists to do.
It has to preserve provenance. Every statement carries its source, the date it was read and how much trust the source has earned. That is what Congregation holds as a native property, not an added field.
It has to measure what it knows. A record that is complete and fresh is not the same as one that is partial and old, and a user should be able to tell the difference without asking. That is what ECP is for: making the quality of the context observable.
It has to behave deterministically when evidence goes stale. Our existing source policy sets a maximum age of seven days for regulatory status and twenty four hours for recalls, withdrawals and safety alerts. When evidence passes its age limit, the system fails closed. Anything that depended on it is paused, and restoring it requires a new reviewed decision, never an automatic one.
Rules that make it usable
An intelligence layer is only worth using if people can see the rules it runs on. Ours are short, and they are written down.
Payment never changes whether an organisation is included or how it is ranked. Organisation evidence and comparison are free to search. Paid tiers cover deeper material such as standalone procurement and geographic records, and they sit on top of the same ranking, not beneath it.
Where a regulatory position conflicts with a commercial one, the regulatory position wins. In the home testing work this is explicit: regulatory evidence is resolved first, then whether a product is suitable for a layperson or restricted to professionals, and only then publication, ranking and commercial handling. Admissibility overrides payment, every time.
Where a source's reuse rights are not recorded, its content is not used. Where a relevance gate decides what belongs, it is measured against thresholds set before the results are seen, so the gate cannot be tuned to flatter itself. We are also careful that it is not so strict that small specialist companies are quietly dropped, because they are often the ones a newcomer most needs to find.
None of this is exciting. It is the part that makes the rest of it believable.
Where people still matter
An intelligence layer does not replace the people who do market entry for a living. It changes where their time goes.
Strategy, positioning and market access consultancy are delivered by partners, not by the software. We keep that list short on purpose: two lead partners for larger companies and one for smaller ones. What the layer does is make their first conversation shorter. Instead of spending a fortnight establishing who the buyers are and what the landscape looks like, they start from an evidenced picture and spend the time on judgement.
Startups and established companies
The two groups need different things from the same layer.
A startup needs speed and a first map. Who buys, through which route, and what is already in place. It cannot afford a month of desk research or a wrong guess about where to begin.
An established company already has a market access team and a long list of territories. It needs breadth, consistency and the ability to compare one market against another on the same terms, with the evidence kept current rather than refreshed once a year.
The UK, with the NHS as the reference market, is where the work starts. International coverage is being added in stages, beginning with the identifiers that let organisations be matched across borders. We would rather widen carefully than claim a coverage we cannot yet stand behind.
The point of the series so far
Every hop has argued the same thing from a different side. Decisions in healthcare are only as defensible as the context underneath them. Market access decisions are no exception. Which market to enter, which route to take, whom to approach first: these are consequential choices made under uncertainty, and they deserve the same standard of evidence as a clinical one.
The best intelligence tools are already out there. What is missing is the layer that lets them agree with each other, say how old they are, and show their working.
What comes next
This hop covers how intelligence tools come together as a governed layer for market access, and the rules that keep it trustworthy.
The next hop goes back to the stack itself. A step by step walkthrough of one clinical scenario, grounded entirely in what the infrastructure does today.