Methodology · v1

The college admissions brain — how it's built.

Every school in this dashboard is built from the same schema, so a coach can move between Hopkins, Brown PLME, and Berkeley without losing their footing — and so we can grow this brain school-by-school without ever needing to redesign the page.

Schools live

100

Private

Public

BS/MD

LAC

Roadmap target

Top 250

Source hierarchy

What we trust, in order.

01The school's own Common Data Set, published by its institutional research office. C1 for accept/yield, C7 for what they value, C9 for SAT/ACT/GPA, C21 for early-decision stats.
02Mission, vision, and strategic plan — pulled from official "About," "President's Office," or "Strategic Initiatives" pages. We never paraphrase the mission; we quote it.
03Admissions blog posts and AO interviews for texture (what an essay should sound like, what an EC list should look like).
04NCES IPEDS as a backstop where a school doesn't publish a recent CDS.

Aggregator sites (US News, College Confidential, Niche) are explicitly avoided for stats.

The schema

Every school record carries the same core shape.

The schema is intentionally narrow: nine quantitative fields, six qualitative narrative fields, three sales-language fields, plus identity, sources, and a last-updated timestamp. New schools are added by appending one record — the dashboard will render correctly the moment the record is saved.

School {
  id, name, type, city, state, logoColor, tags
  cdsYear, acceptanceRate, yieldRate, enrollment
  sat { ebrw, math }, act { composite }
  gpa { note }, testPolicy
  decisionPlans[]              // ED / EA / REA / RD
  academicFactors[]            // CDS C7
  nonAcademicFactors[]         // CDS C7
  mission, visionThemes[]
  whatTheyReallyValue[]        // the coach's translation
  applicationTargets[]         // what a strong app signals
  redFlags[]                   // quiet disqualifiers
  programStrengths[]           // signature majors / institutes
  bsmd?                        // accelerated medical detail
  salesAngle { forParent, forStudent, differentiator }
  sources[], lastUpdated
}

The CDS factor weights

What the four buckets really mean.

Section C7 of the Common Data Set asks every school to rate 19 admissions factors as Very Important, Important, Considered, or Not Considered. This is the single most underused dataset in college admissions — a coach who can read it accurately is immediately more credible than one who can't.

Very Important. The school is publicly committing that this factor can move an applicant from a deny to an admit. Treat it as table stakes.
Important. A meaningful tilt, but not a tiebreaker. Often where "fit" lives.
Considered. Read, but rarely outcome-changing on its own. Useful for context.
Not Considered. Will not be evaluated. Coaches should not over-invest student time here.

How we expand the brain

From 100 to 250.

The transformer at /transform_to_ts.py consumes a single research JSON and emits the typed school array. Adding a school is a three-step ritual: research it (we have a parallelized prompt for this), drop the result into the JSON, regenerate. Page templates do not change. The "living brain" promise is mechanical, not aspirational — every CDS cycle, we re-run, we re-publish.