A List Without the Patients
A business development lead at a company that sells into Life Sciences opens next quarter’s target list. It has the sponsors, the programs, the sites, and a named contact for each. Then a request arrives for a study with narrow eligibility criteria. The sponsor asks one question: where will the team find the patients? The list cannot answer that. It shows who exists, not where eligible patients are, who treats them, or who already recruits them.
A directory describes organizations: who runs a program, who holds which role, and which site has which equipment. Patients are not part of that record, so the list cannot rank a site by the patients it could enroll. It can rank a site only by size and specialty.
So the team uses the usual method. It asks each candidate site to check its own records and estimate how many patients fit. The replies come back, someone totals them in a spreadsheet, and the total clears the sponsor’s target. The proposal says the study is feasible. Each site answers in good faith, but a site cannot see the other studies recruiting the same patients. Some estimates hold up. Others do not, and no one can tell which until the work is committed.
What a Wrong Estimate Costs
The proposal wins, and sites are selected, contracted, and activated. Each one can cost tens of thousands of dollars to open, and some cost more than $100,000. Each then stays open for years while the study enrolls. Industry estimates put the share of sites that never enroll, or fall short of target, near 40%. The range is wide.
A wrong estimate can miss in either direction. A site that overstates its patients earns a place in the proposal and then fails to deliver. A site that understates its patients never makes the shortlist, and no one learns what the team passed over. A study that enrolls late delays the sponsor’s timeline, and a late study weakens the company’s next proposal. Then the team asks the question the list could never answer: where were the patients?
A Faster List Still Has No Patients
Teams at services firms, diagnostics companies, and digital health companies describe the same problem. Their lists come from directories and legacy tools that describe organizations, so targeting rests on assumptions about demand. AI has made those lists faster to build, and many teams now build their own tools with a general model. The model ranks sites in seconds, on whatever it is given. A directory gives it organizations, not patients.
Komodo Health® built Marmot™, an AI platform for healthcare analytics, on Komodo’s Healthcare Map®. The Map captures more than 330 million de-identified U.S. patient journeys. Marmot answers from that claims-based foundation. It is designed around five principles of AI trust: Grounded, Explainable, Reproducible, Verified, and Governed. Every output is transparent and auditable, so a team can see how an answer was built. Marmot also draws on public trial registries, so a team can see where other studies already recruit.
The Pipeline Starts With the Patients
Now run the same pursuit with that visibility. Before the proposal goes out, the team asks where eligible patients are, who treats them, and who already recruits them. It ranks candidate sites on the answers, and the sponsor can check the evidence behind the ranking. A total of site self-reports is harder to check. For revenue leaders, Marmot’s Business Development use case answers those questions. The proposal can then say where the patients are and show how it knows.
