A credit score is a number produced by a statistical model trained to rank borrowers by the likelihood of falling seriously behind. Its component weights reflect predictive power rather than any judgment about behavior.
The model predicts one specific event
Scoring models are built by observing large samples of credit files and recording which accounts later became seriously delinquent within a defined horizon.
Variables that separated those groups are retained and weighted according to how much they contributed. The score is a ranking within the sampled population, not a measure of financial health.
This is why income, savings and employment do not appear. They were not in the credit file the model was built from, whatever their real-world relevance.
Payment history carries the largest weight
Whether obligations were paid as agreed is consistently the most predictive category. Recency, severity and frequency of missed payments each matter, with recent and severe items weighing most.
A single late payment reported by a creditor can affect a score substantially, and its influence fades as it ages while remaining on the report.
Collections, charge-offs, and public record items sit in this category as well, since they represent the outcome the model is trying to anticipate.
Amounts owed is measured in ratios, not dollars
The second largest category considers balances relative to limits on revolving accounts, both individually and in aggregate. Absolute dollar amounts matter far less than the proportion.
Because the figure typically comes from a statement balance reported once per cycle, the reported ratio can differ from what the cardholder considers their usage.
This category responds quickly, since balances change monthly, which is why it moves scores faster than payment history does.
Age, mix and new credit fill the remainder
Length of credit history considers the age of the oldest account and the average age across accounts. Closing an old account can shorten that average once it eventually leaves the report.
Credit mix looks at whether a file contains both revolving and installment accounts, contributing a small amount. New credit counts recent inquiries and newly opened accounts.
These categories together carry less weight than either of the first two, which is why advice focused on them produces limited movement.
There is no single score
Multiple developers publish models, each in several versions, and lenders choose which to use. Industry-specific variants weight the same file differently for auto lending or card issuing.
Scores also differ by bureau, because the three national bureaus hold different data depending on which creditors report to each.
A consumer seeing different numbers from different sources is usually seeing different models over different data, not an error in any of them.