The Completeness Constraint and the Compliance Corridor: Why Applicant-Side Dossier Quality and Regulatory Perimeter, Not Authority Speed, Determine Service Design in German Foreign-Qualification Recognition

This paper tests the dominant assumption in German debate on the recognition of foreign professional qualifications — that administrative slowness is the principal obstacle — against official federal and Land statistics, and finds it unsupported. In Baden-Württemberg the median time to a first decision was twenty days in 2024, and the central recognition body for chamber occupations averaged sixty-nine days from complete submission. The binding constraints lie upstream, in applicant-side dossier completeness, which alone starts the statutory clock under §§ 6 and 13 BQFG, and downstream, in the compensatory-measure tail, where roughly forty-five per cent of decided federal cases land and where the national average interval from order to final decision reached 490 days by 2021. A further friction is the filing channel: the authority handling approximately sixty per cent of Baden-Württemberg procedures requires exclusively postal submission, including expiry-dated originals valid for three months, while the chamber route accepts colour copies and electronic files. The paper then argues that the Rechtsdienstleistungsgesetz, the Zahlungsdiensteaufsichtsgesetz, the Geldwäschegesetz, the General Data Protection Regulation and the Artificial Intelligence Act jointly define a narrow "compliance corridor" determining which remedies a private intermediary may lawfully supply, and that this corridor — not market demand — shapes viable service design. Within it the tractable remedy is a deterministic, versioned Policy-as-Code rule layer combined with statutorily permitted funding advice under § 5(2) no. 3 RDG, rather than paid legal advice or custody of official fees. All statistics are drawn from published official sources — Destatis, Statistisches Landesamt Baden-Württemberg (GENESIS table 21231-0001), BIBB and Landtag von Baden-Württemberg Drucksache 17/9330 — and not from microdata; no new data were generated and no personal data were used. Evidence was collected through an eight-model independent-replication design in which eight frontier language models conducted independent retrieval under an identical brief; conflicting figures are reported as conflicts rather than silently reconciled. Retrieval dates: 30 August 2026 and 12 September 2026. AI-use disclosure: the evidence base was assembled with substantial artificial-intelligence assistance, and language models were also used in drafting and editing. The author verified every substantive claim against the cited primary sources and is solely responsible for the content. Competing interest: the author is the founder of, and an advisory consultant to, Kestura UG (haftungsbeschränkt), which has a commercial interest in the intermediary model analysed, and consults independently through DigiAssist. He is not the company's managing director. The paper's central normative conclusion forecloses the services with the highest apparent commercial value.