What a credit score is counting

A credit score is a number sold to lenders as a compact guess: how likely is this person, compared with other people, to go seriously late on a debt. It is not a report card. It does not know your income, your savings, or whether the debt was a good idea.

The report and the score are different objects. A credit report, from a company such as Equifax, Experian, or TransUnion in the United States, is a file: accounts, limits, balances, and whether payments arrived on time. A score is a model run on that file. FICO and VantageScore are two families of models. A lender can also use its own. When someone says “my credit score” as if there were one, they are compressing several files and several models into a single mood. The number on a bank’s app and the number a car dealer pulls may not match, and both can be real.

FICO has published the rough weights of its classic model, and they are worth knowing as weights, not as a recipe you can game with precision. Payment history is the largest share, about 35 percent. Amounts owed, including how much of your available credit you are using, is about 30 percent. Length of history is about 15 percent. New credit is about 10 percent. The mix of types of credit is about 10 percent. VantageScore uses similar ingredients and does not promise the same percentages. If a website gives you a to-do list that treats these as exact levers, it is rounding a model into a slogan.

Payment history is the record of whether the minimums arrived, not whether you paid the balance in full. A card you pay off every month and a card you revolve can both show “on time.” Thirty days late is the usual first scar that matters. A single late payment does not permanently define a file, but it weighs more while it is recent, and it stays on the report for years. Paying a collection does not make the history unread. The model can see that the account went bad and that it was later settled. Those are two facts, not a deletion.

Amounts owed is where people most often misunderstand the number. The model looks hard at utilization: the balance relative to the limit, especially on revolving accounts. A person with a $400 balance on a $500 limit can look riskier than a person with a $4,000 balance on a $20,000 limit, even though the second person owes more dollars. The model is not measuring virtue. It is measuring how close the available credit is to being used up, which in the historical data has been a sign that a borrower is under strain. Paying the balance down before a statement closes can change the number the model sees. It does not change whether you could afford the spending.