Everything we'd want a friend to ask before paying — what the score means, how refunds work, what we do with your data, and the limits of the model. If we missed yours, the contact desk goes to a real inbox.
We are not an admissions oracle. We are a benchmarking instrument. The difference matters more than people think.
No, and anyone who claims to is selling something worse than this. Admissions is a committee process with too many degrees of randomness to predict. We score where you sit relative to admits, and tell you which levers actually move that.
Models do the pattern-matching, but every feature is benchmarked on real anonymized admit data. We publish error bars on each score and flag low-confidence reads explicitly. No false certainty.
The CDS gives you the median admit's GPA. We model 42 features, including narrative coherence and EC tiering — the things that decide your file in the actual reader's room.
A four-minute structured intake — stats, course list, activities, awards, demographics, and one optional essay sample. No credit card on the Free tier. Applicants are hashed at intake.
Pay once per profile. No subscription, no auto-renewal, no upsell to consulting. Refunds are honest.
Most users run one or two profiles in their cycle. A subscription would punish that. Pay once, get the report, leave.
Never. The composite score, percentile, strengths, and weakness are visible on Free with no card required. The sample report shows exactly what you get.
Premium includes one free re-run within 30 days of purchase — finish that EC, retake the test, then re-score against the same cohort.
No. Free is its own report — a real composite, a real percentile, a real diagnosis. It just stops short of the levers (ranked ECs, narrative dissection, school-specific weaknesses, roadmap) that take serious modeling to produce.
Reports are non-refundable once generation begins, given the computational cost of producing them. Refunds are issued in three cases only: failed report generation, duplicate charges, or technical malfunction. Email billing with your order ID.
Yes — counselor 3-packs, practice 10-packs, cohort 25-packs, and quoted school bundles. See bulk pricing for the per-report ladder.
Forty-two features, benchmarked on actual admits. Every score carries an error bar; low-confidence reads are flagged.
500+ verified accepted profiles, contributed voluntarily and anonymized at intake, plus public sources (Common Data Sets, IPEDS) for baselines. Cohort sizes for school-specific fits are disclosed on the Premium PDF.
Adding features past the point where they have independent signal makes scores look precise without making them more accurate. Every feature in the schema has been ablation-tested for marginal predictive value against the cohort.
A measure of how legibly your profile tells one story to a committee reader. High-spike files compound across activities, awards, and essays toward a single domain; low-spike files read as a checklist of unrelated achievements.
The methodology version is shown on every report (currently v3.0 · Cycle 2026). When the model retrains or the schema changes, the version bumps and old responses remain reproducible for at least eighteen months.
Your file is hashed before it reaches the model. We don't sell data. We don't share intake content. Sub-processors are listed in the privacy notice.
Names, addresses, phone numbers, schools-of-origin, and free-text identifiers are stripped or hashed before scoring. The applicant record is keyed by an opaque hash; the model never reads identifiers. See the privacy notice.
Yes. Email the privacy desk with subject data request and we export or delete within seven days. Deletion is irreversible and removes both the anonymized scoring record and the credential.
No. We do not sell data, ever. We do not share intake content with third parties for advertising or training. Sub-processors (Clerk, Vercel, Stripe, model inference providers) are listed in the privacy notice.
The PDF is generated from the anonymized record. Anything personally identifying is something you typed in yourself.
Bulk packs are self-serve up to 25 reports. Department-scale rollouts are quoted with a data-access agreement. The institutional API is in private development.
Yes. Self-serve 3-, 10-, and 25-packs, plus quoted school and district bundles. Per-report rates ladder from $10 down to $6. See bulk pricing.
Yes — affiliate revenue share, co-produced cohort reads, and sponsored composites for admissions creators. Standards and formats are on creator partnerships.
Programmatic access is in private beta — endpoints, principles, and the waitlist live on data access. Aggregate cohort statistics only; never individual records.
A solo indie team. We don't sell consulting. The disclosures page covers independence and conflicts of interest.
A real person reads every message. No ticket maze, no chatbot, no escalation tier — just the desk that fits your question.