Credit scoring

A credit scoring system is an automated, statistical model that ranks the credit risk of a borrower against other borrowers and sums it up in one number, as the US Federal Reserve explains in its report on credit scoring. Such systems began in the late 1950s at the credit departments of large retailers and finance companies. Inputs come from the borrower's credit file: open accounts, repayment history and the amount of debt.

Banks are not the only users: telecom companies, insurers and landlords check scores too. Newer models add bank account data, which a lender can read through an account information service, and machine learning methods described on machine learning in finance.

On December 7, 2023, the Court of Justice of the European Union ruled in case C-634/21 that SCHUFA's scoring is an automated individual decision, which the GDPR prohibits in principle, as far as clients such as banks give the score a determining role when they grant credit. Whether German data protection law contains a valid exception was left to the Administrative Court of Wiesbaden.

In a parallel judgment the court held that a credit agency may not keep information on a discharge of residual debt longer than the German public insolvency register, which keeps it for six months. The page on lending in Germany covers what the ruling means for lenders.

The new Consumer Credit Directive, Directive (EU) 2023/2225, applies from November 20, 2026. Under Article 18, a creditworthiness assessment may rest on income, expenses and financial obligations, but not on special categories of personal data such as health data, and not on information from social networks. If the assessment is automated, the consumer can ask for human intervention, a meaningful explanation of the main variables and the logic, and a review of the decision.

The EU AI Act lists AI systems that evaluate the creditworthiness of natural persons or establish their credit score as high-risk, with an exception for fraud detection. A bank that uses such a model has to meet the act's duties on data, documentation, human oversight and risk management, covered on the EU AI Act in financial services.

Finance Loop is the meeting place for credit risk teams, data scientists, lending fintechs and the compliance staff who have to explain a scoring model to a supervisor. It connects the finance, IT and AI communities in Germany, Austria and Switzerland, with events in Frankfurt, Munich, Berlin and Hamburg.

Finance Loop announced KI Exchange 2026 in Hamburg, a Payment & Banking conference on AI as infrastructure for banks, with governance and fraud detection on the program. It is a partner of the FinTech Founder & Investor Evening in Frankfurt, where the CEO of the SME lender Teylor speaks.

Credit scoring is an automated statistical method that predicts the probability that a borrower repays a loan, based on data such as repayment history, open accounts and debt. The result is a single score that lenders use to decide whether to lend and at what price.

Yes, under conditions. The EU Court of Justice treats a score as an automated decision when a bank relies on it decisively, which the GDPR allows only with a legal basis and safeguards. Scoring continues, and the consumer has the right to meaningful information about the logic involved.

Yes. Annex III of the AI Act lists AI systems that assess the creditworthiness of natural persons or set their credit score as high-risk. Systems used only to detect financial fraud are excepted.