What Actually Affects Your Credit Score

Search for how credit scores work and you will find the same pie chart everywhere: 35% payment history, 30% amounts owed, and so on. Those figures come from one scoring company's general guidance about one family of models. They are not a rule, and they are not published by any regulator.

Here is what is actually established.

There is no single score

The CFPB describes a credit score as a prediction of your credit behaviour, such as how likely you are to repay a loan on time. Crucially, scores vary depending on the scoring model used, which credit bureau's data it is run against, and the date it was calculated.

That means the number your card issuer shows you free each month may differ from the one a mortgage lender pulls, and neither is wrong. Most scores fall in a 300 to 850 range, with higher scores generally producing better approval odds and rates.

Chasing a specific number across different providers is a waste of energy. Watching the direction of travel is not.

The factors that scoring models consider

Per the CFPB, credit scoring models typically take account of:

Payment history - whether you have paid on time
Current unpaid debt - how much you owe now
Number and type of loan accounts you hold
Length of time you have held your accounts
Credit utilisation - how much of your available credit you are using
Recent applications for new credit
Negative events such as collections, foreclosure or bankruptcy, and how recently they occurred

Notice what the regulator does not do: assign weights. Different models weight these differently, which is precisely why your scores differ between providers.

What this means in practice

Payment history is the one nobody disputes. Every model considers it and a missed payment is the most damaging single event most people will experience. Automating at least the minimum payment on every account protects against the worst outcome, even in a bad month.

Utilisation is the fastest lever. It is calculated from your reported balances against your limits, and it updates every cycle. Unlike payment history, which takes years to rebuild, utilisation can change within a month. Paying a card down before the statement date - not just before the due date - is what reduces the balance that gets reported.

Age of accounts rewards patience and punishes closures. Closing an old card you no longer use removes both its history and its credit limit, which can raise your utilisation on everything else. If it carries no annual fee, leaving it open and occasionally used is usually the better call.

Applications matter, but less than people fear. A single application has a modest, temporary effect. Several in a short window is a different signal.

What is not in there

Your income is not a factor in your credit score. Neither is your savings balance, your job, or your net worth. Lenders consider those separately when they assess an application, but they are not inputs to the score itself.

Checking your own score does not affect it either.

A reasonable approach

You do not need to optimise a number you cannot see the formula for. You need to:

Never miss a payment
Keep reported balances well below your limits
Open new accounts deliberately rather than opportunistically
Leave old, fee-free accounts open
Check your credit reports for errors, which you are entitled to do free

Errors are more common than people assume, and a wrong entry can cost you more than any amount of optimisation gains you.

How long negative information stays

The duration matters as much as the event, and it is one of the few parts of the system with firm, published timeframes rather than model-specific guesswork.

Item	Generally stays on your report for
Late payments	7 years
Collections accounts	7 years from the original delinquency
Chapter 13 bankruptcy	7 years
Chapter 7 bankruptcy	10 years
Hard inquiries	2 years, though typically only scored for about 1
Closed accounts in good standing	Up to 10 years

Two things follow from this. First, the damage from a single missed payment is real but finite, and its weight fades as it ages rather than staying constant until it drops off. Second, a closed account in good standing keeps contributing history for years, which is why closing an old card is less immediately destructive to your account age than people fear, and more destructive to your utilisation, which is covered below.

Where the effort actually pays

Not every factor responds to effort on the same timescale. Ranking them by how quickly you can move them is more useful than ranking them by weight.

Factor	Time to change	What moves it
Credit utilisation	One billing cycle	Paying down balances before the statement closes
Recent applications	Months	Simply not applying
Payment history	Years	Time, plus never adding another miss
Length of history	Years	Time only; nothing you do accelerates it
Account mix	Slow, and rarely worth engineering	Ordinary borrowing over time

The practical reading: if you need a better score for something specific in the next few months, utilisation is essentially the only lever that responds in time. Everything else is a matter of not making things worse and waiting.

A worked example of the utilisation lever

Suppose you hold three cards with a combined limit of $12,000 and balances totalling $6,400. Your overall utilisation is about 53%.

You have $2,000 available to put against the balances. Where it goes changes the outcome:

Spread evenly across the three, your overall figure falls to about 37%, and each card individually improves a little.
Concentrated on the card that is nearly maxed, your overall figure falls the same amount, but the per-card figure on the worst offender improves sharply.

Because models look at both per-card and overall utilisation, clearing the card closest to its limit generally does more than spreading the same money thinly. See credit utilisation explained for how the reporting snapshot decides which balance is actually seen.

Your reports and your score are different things

Your credit report is the underlying record: accounts, balances, payment history, inquiries, public records. Your credit score is a number a model produces from that record. You are entitled to the report; the score is a product.

This distinction matters because errors live in the report, and an error there propagates into every score calculated from it. Checking reports is therefore higher value than checking scores, and it is free at AnnualCreditReport.com, the site established under federal law for this purpose.

What to look for, in order of how much damage it does:

Accounts you do not recognise, which can indicate identity theft
Payments marked late that you made on time
Balances or limits that are wrong, since a limit reported too low inflates your utilisation
Debts appearing twice, for instance both as an original account and a collection
Negative items older than the periods in the table above, which should have dropped off

If you find an error, you have the right to dispute it with the bureau, which must generally investigate.

What genuinely does not affect it

This list is worth knowing because a large amount of anxiety attaches to things that are simply not inputs:

Your income, savings, job, or net worth. Lenders weigh these when assessing an application, but they are not in the score.
Checking your own score or report. This is a soft inquiry and is not scored.
Your age, marital status, race, religion, or where you live. Using several of these in lending decisions is prohibited outright.
Paying with a debit card or cash, which simply produces no credit data either way.
Carrying a small balance to show activity. This is a persistent myth. Paying in full is not penalised, and carrying a balance costs you interest for no scoring benefit.

If your score is already damaged

The honest answer is that recovery is mostly time plus not repeating the event, and anyone selling faster is selling something. That said, the sequence that works is:

Stop the bleeding. Get every account current, because an account that is currently late keeps doing damage in a way an old, cured late payment does not.
Automate at least the minimum on everything, so a bad month cannot become a missed payment.
Bring utilisation down, since it is the fastest-responding factor and carries no waiting period.
Stop applying for new credit for a while.
Pull your reports and dispute anything wrong, which is the only route to a fast improvement that is legitimate.
Then wait. Negative items fade in influence well before they drop off entirely.

Credit repair companies cannot lawfully remove accurate negative information, whatever they imply. Anything they can legitimately do about inaccurate information, you can do yourself for free.

Different scores, and why they disagree

Being told you have "a" credit score is misleading. You have many, and the spread between them can be substantial.

Source of difference	Effect
Scoring model and version	Different factors, different weights
Which bureau's data	Not all lenders report to all three
Date calculated	Balances change every cycle
Industry-specific versions	Auto and card versions weight differently

A free score from a card issuer and the score a mortgage lender pulls are often different models entirely, and a gap of several tens of points between them is unremarkable rather than an error.

The useful conclusion is to stop treating any single number as your score. Watch one source consistently and track its direction over months. Direction is real information; the absolute figure from any one provider is not.

Sources

Current as of August 2026.