Methodology
Shortlist is built on a simple invariant: every fact you see should be traceable to a source, dated, and labeled with a confidence class. Every score should be able to explain its inputs.
Source priority
- Official federal datasets (IPEDS, College Scorecard, NSF HERD, Clery).
- Official institution-published pages and PDFs.
- State higher-ed agencies and accreditors.
- Public local/geospatial datasets for amenities and context.
- Editorial review by the Shortlist team where structured sources do not cover a question.
- User-submitted corrections, reviewed by the team before they affect public facts.
Advertiser-supplied claims are kept in a separate ledger and never enter the organic fact graph until they are independently verified.
Confidence classes
- High — official structured source or manually verified institution page.
- Medium — official unstructured source parsed with review, reputable third-party source, or cross-source agreement.
- Low — single scraped mention, stale page, or user contribution pending review.
- Unknown — no reliable data is available; we say so explicitly rather than guessing.
Fit ranking
Fit ranking is deterministic: the same inputs always produce the same output, including the explanation. We score five dimensions — academic, financial, practical, outcomes, and cultural — and combine them with the user's weights.
- Hard filters (state, distance, max net price, size) eliminate candidates outright.
- Soft preferences move the score up or down within each dimension.
- Missing values are treated as unknown, not zero. Confidence is reported separately.
- Sponsored placements never affect organic ranking. The ranking engine does not read sponsorship tables.
We deliberately do not publish a single master rank. Fit is multidimensional, and any single number obscures tradeoffs that families need to see.
SEO and content quality
Public pages are indexable only when they meet a coverage and uniqueness threshold.
Thin or stale pages carry a noindex directive automatically until enough
source-backed data is present.
Privacy and minors
We do not allow self-serve accounts for users under 13 until verifiable parental consent is implemented. Sensitive preferences (identity, support needs, finances) are optional, excluded from advertising, and stored with stricter access controls. See the privacy notice for details.
Corrections
Anyone can report a problem with any fact or page. Corrections enter a queue, are reviewed by an editor, and resolved with an audit trail. Institutions cannot suppress unfavorable factual data.