What a credit score is actually predicting, the five factors FICO weights and by how much, what does not count at all, and the fixes that move the number fastest.
A credit score is not a report card on how responsible you are. The Consumer Financial Protection Bureau defines it as something narrower and stranger: a prediction of your credit behaviour, such as how likely you are to pay a loan back on time, generated from the information in your credit reports. That distinction matters more than it sounds. A score is not measuring your virtue, your income or your savings. It is a statistical guess about one specific future behaviour, built from one specific set of records.
Once you understand what the number is trying to predict, the five factors behind it stop looking arbitrary and start looking obvious.
First: you do not have a credit score
You have dozens. The CFPB is explicit that consumers do not have a single credit score, because scores vary depending on the scoring model used, which credit report the data came from, and the date it was calculated.

There are two dominant model families, FICO and VantageScore, each with multiple versions still in active use. There are three national credit reporting companies, Equifax, Experian and TransUnion, and a lender may pull from one, two or all three. A mortgage lender and a credit card issuer will often use different FICO versions entirely. This is why the score in your banking app and the score your lender quotes can differ by thirty points without either being wrong.
Most scores land on a 300 to 850 scale. Chasing the exact number in any one app is a waste of attention. What moves one score generally moves the others, because they are all reading the same underlying reports.
The five factors, and what each one is actually measuring
FICO publishes the weightings its models place on five categories. These are the figures below, and they are the reason the advice that follows is ordered the way it is.
Payment history, 35 percent
Whether you have paid past credit accounts on time. It is the single largest input, and it is the one people underestimate because it seems too simple.
The important detail is that severity, recency and frequency all register separately. A payment thirty days late is a different event from one ninety days late. A late payment from last month weighs more than one from four years ago. One slip in an otherwise clean file reads differently from a pattern.
Worth knowing: most creditors do not report a payment as late until it is a full thirty days past due. Paying five days after the due date will usually cost you a late fee and possibly your promotional interest rate, but it does not typically reach your credit report. That is not permission to pay late. It is a reason not to panic if you missed a date by a few days.
Amounts owed, 30 percent
Not how much debt you have in absolute terms, but how much of your available credit you are using. This is credit utilisation, and it is where most people have the fastest available gains.
The arithmetic is simple. Add up your credit card balances, add up your credit limits, divide the first by the second. A person with 2,000 dollars of balances against 10,000 dollars of limits is at 20 percent utilisation. A person with the same 2,000 dollars against a single 2,500 dollar limit is at 80 percent, and will score materially worse on this factor despite owing identical money.
Scoring models look at utilisation both overall and per card, so one maxed card can drag you even when your total looks healthy:
| Situation | Total balances | Total limits | Overall utilisation | Worst single card |
|---|---|---|---|---|
| Spread across three cards | 3,000 | 15,000 | 20 percent | 20 percent |
| Same debt, one card carrying it | 3,000 | 15,000 | 20 percent | 95 percent |
| Less debt, tight limits | 1,800 | 2,500 | 72 percent | 72 percent |
The second row is the trap. Overall utilisation looks fine, and the file still shows a card running at 95 percent.
Two practical consequences follow. First, utilisation carries no memory. It is recalculated from whatever balance your issuer reports each cycle, so paying a card down this month improves the factor next month. Nothing else on this list moves that fast. Second, closing an unused card removes its limit from the denominator and can raise your utilisation without you borrowing a cent.
Length of credit history, 15 percent
FICO looks at the age of your oldest account, the age of your newest, and the average age across all of them.
This is the factor you cannot rush, which is precisely why it deserves defensive attention. The oldest card in your wallet is doing quiet work simply by existing. Closing it does not erase it immediately, since closed accounts in good standing can remain on your report for years, but it stops ageing as an open account and eventually drops off.
If you are considering tidying up your finances by cancelling cards, the no-annual-fee card you have held longest is usually the one to keep, even if you barely use it. A small recurring charge is enough to keep it active.
Credit mix, 10 percent
Your combination of account types: credit cards, retail accounts, installment loans, finance company accounts and mortgages. Models read a borrower who has handled both revolving credit and installment debt as marginally less risky than one who has only ever held a card.
Ten percent is not nothing, but it is nowhere near enough to justify taking on a loan you do not need in order to diversify. Mix improves naturally as life happens: a car loan, a mortgage, a student loan. Manufacturing it is a poor trade.
New credit, 10 percent
Opening several accounts in a short window registers as elevated risk, particularly for people with shorter histories.
A hard inquiry, the kind generated when you formally apply for credit, is visible to lenders. A soft inquiry, including checking your own score, is not scored at all. Checking your own credit cannot hurt it, a myth that costs people more anxiety than any real scoring factor.
Rate shopping is treated sensibly. Multiple inquiries for the same type of loan within a short shopping window are generally grouped as one event, so comparing mortgage or auto lenders is not penalised the way five separate card applications would be.
The percentages are averages, not your file
FICO is clear that these weightings describe the general population, not any individual. The importance of each category shifts with the contents of your report. Someone with a thin file, only a year or two of history, has length and new credit weighted differently from someone with two decades of accounts. FICO also notes that it is not possible to isolate the exact impact of a single factor without looking at the whole profile.
So treat the five numbers as a priority order, which is what they are genuinely useful for, rather than as a formula you can reverse-engineer.
What does not affect your score
Credit reports contain credit information, and scores are built only from what is in them. Not included: your income, your savings or checking balances, your employment status, your age, your marital status, your education, or your race, religion or national origin, which lenders are prohibited from considering.
This explains a situation people find baffling. A person with a large salary and no debt can have a weaker score than someone earning a third as much, because the second person has ten years of consistently paid accounts and the first has almost no credit history to predict from. The model is not judging wealth. It is predicting repayment behaviour, and it can only do that from evidence of repayment behaviour.
How long damage lasts
Under the Fair Credit Reporting Act, the CFPB states that a credit reporting company can generally report most negative information for seven years. Bankruptcies can remain for up to ten. Lawsuits and judgments stay for seven years or until the governing statute of limitations runs out, whichever is longer.
Two things follow. Negative marks are not permanent, and their weight fades well before they fall off, because recency is part of how payment history is assessed. And no company can lawfully remove accurate negative information early. Any service promising to delete correct information from your report is selling you something that does not exist. What you can do, and should, is dispute information that is genuinely inaccurate.
Check yours without paying for it
Since October 2023, the three national credit reporting companies have permanently offered one free report from each, every week, through AnnualCreditReport.com. The Federal Trade Commission warns explicitly that other sites may charge you or may be fraudulent sites set up to steal your personal information. Use that address and no other.
A report is not a score, and this is the more useful of the two. Your report is the raw data everything is calculated from, and errors in it are common enough to be worth checking for: accounts you do not recognise, payments marked late that were not, balances that are wrong, or duplicate collections. Fixing an error is the only route to a score improvement that is both immediate and free.
What to do, in order
Ranked by how much difference each makes, not by how satisfying it feels:
- Automate the minimum payment on everything. Payment history is 35 percent, and automation removes the only failure mode that really matters. Pay more manually when you can.
- Bring down your highest-utilisation card first. Amounts owed is 30 percent and updates every cycle. Attacking the card closest to its limit does more for the score than spreading payments evenly, though paying the highest interest rate first does more for your wallet. They are different goals; know which one you are pursuing.
- Stop closing old cards. It shortens history and shrinks your available credit at the same time, hitting two factors at once.
- Ask for a limit increase rather than a new card. A higher limit improves utilisation without a new account dragging your average age down. Check whether your issuer does this with a soft pull.
- Space out applications, and group rate shopping. Compare mortgage or auto lenders inside a short window; do not scatter card applications across a year.
- Read your reports from all three bureaus. Free, weekly, and the only place you will find errors that are costing you points you never earned away.
What is conspicuously absent from that list is anything clever. There is no technique here, no exploit, no leverage. The score is predicting whether you pay what you owe, on time, without running your available credit to the edge. It mostly succeeds at that, which is why the honest advice is so boring.
Where these figures come from
- Consumer Financial Protection Bureau, What is a credit score?
- myFICO, What is in my FICO Scores?
- Consumer Financial Protection Bureau, How long does information stay on my credit report?
- Federal Trade Commission, You now have permanent access to free weekly credit reports
- AnnualCreditReport.com
General information, current as of September 2026, and not personalised financial advice. Scoring models and the figures above change; check the primary sources before acting on anything with money attached.
Frequently asked questions
How fast can a credit score improve?
Utilization changes can show up within one or two statement cycles, since card balances are reported monthly. Rebuilding after late payments takes longer, often many months of clean history.
Do I have one credit score or many?
Many. Each bureau holds its own report, and different scoring models read them differently, so small differences between scores you see are normal.
Does paying rent build credit?
Only if it is reported. Some services and landlords report rent payments to the bureaus; ask before assuming your on-time rent is counting.

