evidoria

← Back to browse

Good practice Imported

Germany's DoSV — A 14-Year-Old National Algorithm for Matching Students to University Places

Germany · Dortmund · See the Germany profile

Evidence: Observational / pre–post Top 59% 47/100 · Ask Evidence Copilot about this practice

Since winter 2012/13, Germany's DoSV has run a nationwide deferred-acceptance algorithm matching applicants to capacity-restricted degree programmes. In 2024/25 it coordinated 155 universities and 1,762 programmes, including 32,966 applicants for 10,219 medicine places.

155 universities
Universities coordinated through the DoSV (winter 2024/25)
1762 programmes
Study programmes coordinated through the DoSV (winter 2024/25)
32966 applicants
Applicants competing for medicine places (winter 2024/25)
10219 places
Medicine places available (winter 2024/25)
3.2 ratio
Applicants per medicine place (winter 2024/25)
14 years
Years the DoSV has run continuously (2012-2026)

Details

Maturity
Established
Promoter
Stiftung für Hochschulzulassung (Foundation for University Admissions)
Period
Winter semester 2012/2013 – present
Keywords
higher-education admissions, matching algorithms, public administration

Context

Before 2012, German applicants to capacity-restricted ('NC') programmes had to apply separately to each institution, often holding multiple simultaneous offers or none, leaving seats empty even as qualified applicants went unplaced. The Stiftung für Hochschulzulassung (Foundation for University Admissions) built the Dialogorientiertes Serviceverfahren (DoSV), a centrally coordinated, dialogue-based matching process.

Objectives

Let applicants register once via hochschulstart.de, apply to up to 12 programmes, and rank preferences, while institutions independently rank applicants, resolved through an iterative university-proposing deferred-acceptance process that lets applicants hold and re-rank multiple offers before committing.

Activities

For medicine, dentistry, veterinary medicine and pharmacy, admission is split by statute into three quotas: 30% by Abitur grade, 10% by additional qualifications, and 60% by each university's own selection criteria.

Results

For winter semester 2024/25, 155 universities coordinated 1,762 study programmes through the DoSV; within the medicine quota alone, 32,966 applicants competed for 10,219 places, roughly 3.2 applicants per place. A peer-reviewed 2022 Journal of Political Economy study using real DoSV data found applicants respond suboptimally to early offers under simpler mechanisms, supporting the dynamic multi-offer design actually used.

Conclusions

The DoSV is a rule-based optimisation/matching algorithm rather than a machine-learning system; no major scandal or bias finding has been reported against it, with the main documented friction being applicant confusion over its multi-week, multi-quota process.

Implementation

Indicative cost
High (€500k–€5M)
Time to results
Long (> 3 years)
Staffing & skills
Stiftung für Hochschulzulassung (Foundation for University Admissions) staff, Participating universities' admissions offices

Conditions for success

  • Statutory quota framework (30/10/60) giving legal clarity to selection criteria
  • Centralised single-registration portal (hochschulstart.de) reducing applicant friction
  • Iterative deferred-acceptance design allowing applicants to hold and re-rank multiple offers

Common failure modes

  • Applicant confusion over the multi-week, multi-quota process, the main documented friction
  • Each university's internal 60%-quota selection criteria are not independently audited

Where it fits

Governance type
public-law foundation operating a national mechanism
Scale
national (155 universities, 1,762 programmes)
Income level
high-income (Germany)

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful