Data & methodology.

What Circulus is

Circulus combines a structured university catalog (PostgreSQL via Prisma), your profile, shortlist, and application tracker so discovery and execution stay in one workspace. We aim to show what we know, label gaps, and avoid implying precision we do not have.

Catalog depth & “quality tier”

We label institution and programme rows with a simple deep / good / partial signal (shown as “Deep catalog”, “Good coverage”, or “Partial data”) based on whether tuition, deadlines, languages, curriculum topic mix (when ingested), and verification timestamps are present. This is not a ranking of how “good” a university is — only how complete our row is for decisions.

Circulus signals on university pages turn existing fields into short, honest summaries — e.g. teaching-language breadth and selectivity context from acceptance rate and applicant counts when published.

Catalog & freshness

Universities and programmes are ingested and enriched from official and licensed sources where possible. Many rows include verification timestamps and source links on detail pages when available. Rolling refresh is prioritized by traffic and launch-corridor demand — not all entries update on the same cadence.

CI enforces minimum catalog coverage thresholds for releases (see npm run coverage:universities:gate). Incomplete tuition, deadlines, or teaching-language fields trigger a Limited info badge on cards so decisions are not made on silent gaps.

Launch corridors (priority)

We deliberately deepen quality for a few applicant journeys before chasing global breadth:

• Non-EU → EU English-taught Master. Students outside the EU targeting English-taught Master programmes in Northern and Western Europe.
• EU → UK undergraduate. European applicants exploring UK Bachelor routes (UCAS-aligned planning).
• International → DACH German-taught. Applicants preparing for German-taught programmes in DE/AT/CH.

AI assistance

AI features receive your student profile, shortlist counts (Free) or labeled shortlist and application rows (Premium, capped for safety), plus task-specific context (e.g. match scores). Successful responses from match explanation include a Context used summary. AI output is assistive — always verify deadlines and requirements on official university sites.

Daily AI usage limits apply per account tier; premium-only tools are enforced server-side (see Pricing).

What-if simulator (Premium)

The admission simulator estimates a percentage band for schools on your shortlist (applications, favorites, and optional pins) using the same structured match engine as elsewhere, then widens the band when important signals are missing (e.g. personal statement, CV) or when a school is highly selective. The band is a modelled uncertainty range, not a prediction of admission probability.

Not modeled in full: committee judgment, interview performance, reference strength, cohort competitiveness for a given intake year, and portfolio quality. AI narrative and improvement tips are grounded in your profile and our catalog fields; they must not contradict the engine band shown in the product.

Limits and billing: Premium-only; up to 10 schools per simulation and 10 full analysis runs per month (debounced odds preview does not consume this quota). See Pricing.

Report a data error

Signed-in users can use Report data issue on university and programme modals; we route reports to our ops inbox when email is configured. You can also email kontakt@personal-plaza.de with the university name, programme (if any), and a link to an official source when possible. Critical factual errors are triaged with a best-effort response window (target: 72 hours for issues that could mislead applicants).

Privacy & export

Account data export and deletion are available from Settings. See the Privacy Policy for legal bases and processors.