Credit check

What a credit check delivers

A credit check estimates before an order how likely a customer is to pay an invoice. It is therefore the basis for a decision about payment method: instant payment for everyone, invoice purchase only for sufficient reliability. Without such a check the shop carries the default risk in full.

Placement matters. The check does not answer whether a customer will pay, only how high the risk is in statistical terms. Every decision rests on probabilities, and every threshold unavoidably produces two kinds of error: customers wrongly restricted and customers wrongly granted invoice purchase.

How the check runs

• Identity check: Whether name, address and date of birth match. Faults here are the most frequent cause of a wrong rating.
• Historical data check: Whether findings exist about earlier payment problems. For new customers an address check is used instead.
• Scoring: A value is built from the features and compared with a threshold. The result is a class, not a score awarded to the customer.
• Decision: Which payment methods are offered. That decision must be implementable in the shop, otherwise the cart breaks.

What the law requires

The check processes personal data and needs a legal basis. Here that is the legitimate interest in avoiding payment defaults, weighed against the interests of the customer. Analysing address data, credit agency data and your own experience is permissible. Decisions tied to traits such as origin, religion or gender are not, and opaque full automation is not either.

A second point concerns transparency. Affected individuals must be able to learn that an automated check took place, what kind of data was used and how they can contest the outcome. Those duties are best met with a short explanatory paragraph exactly where the payment method is restricted, plus a link to your own privacy notice.

Limits and common mistakes

The limits lie in the method itself. Scoring judges a group, not a person, and it reproduces patterns from the past. Someone living in an area with high tenant turnover can be systematically rated worse without ever having left an invoice unpaid. Anyone deploying the check should know and document that side effect.

The most common implementation mistake is a rule that excludes too many customers. A shop refusing invoice purchase at every small signal loses revenue and customers, because some go straight to a competitor. A tiered logic works better: always offer immediately available payment methods, grant invoice purchase on an inconspicuous check, and cap the amount rather than refusing outright when something looks off. That keeps the customer in the shop and the risk contained.