What a credit file actually records, and what it does not

Everything below about credit files comes from what actually happens rather than from what is supposed to.

The score you see is not the score they use

Credit reference agencies sell consumer-facing scores, and lenders build their own models from the underlying file data. Two agencies can show materially different numbers for the same person because they hold different data and weight it differently. The useful thing is the file itself — accounts, balances, payment history, searches — because that is what lenders actually read.

Treating the consumer score as a target rather than an indicator leads to some genuinely counterproductive behaviour.

What is on it

Typically: credit accounts and their balances, payment history over several years, credit searches, public records such as judgments, and electoral roll data where applicable. What is generally not on it: income, savings, employment, and in most systems, everyday transactions. This surprises people, because it means a high income does not improve a file directly.

What counts as reportable also varies by country: rent, telecoms and utility payments feed some national systems and are absent from others, which is why advice about building a file rarely transfers across a border.

Utilisation and history do most of the work

Consistently paying on time is the largest positive factor in almost every model. Using a high proportion of available credit is a strong negative signal, and the effect appears well below the limit.

Practically, keeping balances low relative to limits therefore helps even when you clear them in full each month, because the reported balance is usually the statement one. Since that figure is a snapshot taken on one day, paying part of the balance before the statement date rather than after it changes the number a lender sees without changing anything about what you spent.

Common misconceptions

Checking your own file is a soft search and has no effect; only applications create hard searches. Closing old accounts can reduce your available credit and shorten average account age, which sometimes makes things worse rather than better. Not borrowing at all does not produce a strong file, because there is nothing to demonstrate repayment behaviour.

There is also no shared blacklist and no record of the refusal itself: the next lender sees the search rather than the decision, and reaches its own conclusion from the same underlying data.

Fixing errors is worth the effort

Files contain mistakes with some regularity — accounts that were closed, defaults that were settled, addresses that do not exist. Every jurisdiction with credit reporting has a dispute process, and agencies are obliged to investigate.

Correcting a single erroneous default can change borrowing outcomes far more than months of careful behaviour. Where a dispute is not resolved in your favour, many systems allow a short statement of correction to be attached to the entry, which a human reading a manual referral will see and an automated score will ignore entirely.

Rates, thresholds and rules differ by country and change often — check current figures before acting.

Associations and addresses link one file to another

A joint account or joint borrowing generally creates a financial association, which means another person's record can be read alongside yours even when you apply alone. That link commonly outlives the relationship, and in most systems it has to be broken by a separate request once the joint borrowing has actually closed.

For most households, moving country usually resets the file to nothing, because reporting is organised nationally and history rarely crosses a border with you. Lenders also assess affordability from declared income and outgoings, which is why a clean file and a refused application are not a contradiction.

The takeaway

Read the file, not the score. The file is what a lender is actually looking at.

Write the number down before you decide. It usually decides for you.