We will not predictwhere you get in.

Published rankings sort medical schools by research funding, reputation surveys and selectivity. Those are real attributes of real institutions, and they produce exactly one list. It is the same list for every applicant in the country.

That list is useful if your question is which school is most prestigious. It is close to useless if your question is which school wants an applicant who spent four years in community health and intends to go back to the county they came from. That is a question about a relationship between two things, and a ranking only measures one of them.

It would be easy to ship a score and stop. A number sorts, and sorting feels like an answer. But a number you cannot interrogate is a number you cannot act on, and an unexplained score tells an applicant nothing they can use.

So every school on a PairScore list carries a written account of the reasoning: which experiences the model read, which part of that school's stated mission it matched them to, and what it could not find. The last of those is the most useful thing on the page.

A Position Tier is derived from your PairScore for that one school. It describes the strength of an applicant–school relationship rather than functioning as an admissions probability or an applicant percentile, and it says nothing about where you sit relative to other applicants.

Low Fit is also not red anywhere in this product. A school that does not align with you today is not a rejection and not a failure, and coloring it like one would be a design decision that made people feel worse for no informational gain.

PairScore tells an applicant where they align. The same framework, applied in partnership with To Be An M.D., gets them in front of someone who has already done it: a medical student or a physician, matched on shared experience themes rather than on alphabetical order.

Programs running that second half have been doing it in spreadsheets, spending a week of a coordinator's time on an assignment problem that a stable-matching pass solves in one run. Both products exist because the same student is failed twice: once by a list that was not built for them, and once by a mentorship pairing that was made in a hurry.