A credit file records behaviour, not character or income

What the file is for

A lender deciding whether to advance money faces a problem it cannot solve from the application alone: it needs some evidence about how the applicant has handled obligations before. Credit reference agencies exist to supply that evidence. They collect information from lenders and from public records, assemble it into a file, and sell access back to lenders assessing an application. The file isn’t a verdict; it is raw material.

The file is therefore a history of credit behaviour rather than a judgement of a person. It contains the accounts held, the balances, the payment record on each, how long each has existed, recent applications, and certain matters of public record such as court judgments or insolvency, where the local system provides for that.

The exact contents, the retention periods and even the number of agencies differ by country. What is fairly universal is the underlying logic: past handling of obligations is the best cheap predictor available of future handling of them.

The things that are usually absent

Several items people expect to find on a credit file are not on it in most systems. Income is generally not recorded, which surprises almost everyone; a lender learns it from the application and from documents, not from the file. Savings balances are usually absent. So, typically, is anything about employment beyond what an applicant declares.

The absence of income has a practical consequence worth stating. A high earner with a poor payment record and a modest earner with a clean one look quite different in the file, and it is the record rather than the earnings that the file speaks to. Affordability is assessed separately, from other information.

This also means a household can be entirely solvent, comfortable and careful, and still present a thin file if it has simply never borrowed. Absence of evidence isn’t the same as good evidence, which is why a lender may hesitate over someone who has done nothing wrong.

Scores are produced by the reader, not stored in the file

Consumers are shown scores, and it is easy to conclude that a single number sits on the file waiting to be consulted. Generally it does not. The file holds data; a score is calculated from that data by whoever is doing the assessing, using their own model and their own weightings, which they do not publish.

That is why two lenders can reach different conclusions about the same applicant on the same day, and why the score shown by a consumer service is an indication rather than the figure a lender sees. It is a useful proxy for direction — improving or deteriorating — and a poor guide to any specific decision.

It also means chasing a score as a target can mislead. The underlying behaviour is what the models are reading, and behaviour that looks good to one model may matter less to another.

What tends to move the assessment

Across most systems, a few factors do most of the work. Payment history is the heaviest, because it speaks most directly to the question being asked. How much of the available credit is being used matters, since a facility running near its limit reads differently from the same balance on a larger one. Length of history helps, which is why closing a long-standing account is not always the tidying-up it appears to be.

Frequent applications in a short period tend to weigh against an applicant, because a burst of applications is a signal in itself. Many systems distinguish between a full application and a preliminary check that does not leave the same mark, though the mechanics vary.

What does not appear, in most systems, is any record of ordinary spending, of who you live with beyond a formal financial link, or of anything a lender is not permitted to consider. Financial associations created by joint borrowing are a real and often overlooked exception, since they can connect two files.

Reading your own file, and correcting it

Most jurisdictions give people a right to see their own file and to have inaccurate entries corrected. That right is worth exercising periodically, because errors do occur — accounts wrongly attributed, settled debts still showing as open, addresses conflated — and they are considerably easier to fix in advance of an application than during one.

Correction is a process rather than a request: the agency generally has to investigate with the lender that supplied the data, and there is usually a mechanism for adding a short explanatory note where a disputed entry is factually correct but incomplete.

What no legitimate service can do is remove accurate adverse information before its retention period expires. Offers to do so should be treated with the suspicion the claim deserves. Since the rules, agencies and rights differ substantially between countries, and since a refused application has consequences of its own, anyone facing a significant borrowing decision is better off checking the position locally and, where the sum is large, taking regulated advice.