Credit Score Explained: How Cards Affect Your Score

A credit score is not a mystery. It is a formula. The inputs are publicly known, the weightings are published, and the outputs are predictable if you understand the mechanics. What feels opaque becomes straightforward once you see the calculation.

The regulator's own definition is deliberately unglamorous. A credit score, says the Consumer Financial Protection Bureau, is "a prediction of your credit behavior, such as how likely you are to pay a loan back on time, based on information from your credit reports." It is a risk estimate, not a report card, and it exists for the lender's benefit rather than yours.

One correction before the mechanics, because it saves a lot of confusion later: you do not have a credit score. The CFPB is blunt about this — "You do not have just 'one' credit score. Each credit score depends on the data used to calculate it, and it may differ depending on the scoring model... the source of the data used, and even the day when it was calculated." Different bureaux hold different data, different lenders buy different models, and the number your banking app shows you is often not the number the underwriter sees. Chasing a single figure to the point is wasted effort. Chasing the underlying behaviour is not.

Here is exactly how the calculation works and what credit cards do to each factor.

The Five Factors (FICO Model)

The most widely used scoring model — FICO — calculates your score from five factors with published weightings. Those weightings are averages, not laws: FICO states that "the levels of importance shown in the FICO Scores chart above are for the general population and may be different for different credit profiles." If your file is thin, the mix shifts.

Payment History — 35%

The most important factor by a significant margin. Every on-time payment adds a positive mark. Every late payment reported at 30 days past due adds a negative mark, and it is durable: the CFPB confirms that a credit reporting company "generally can report most negative information for seven years," with bankruptcies staying up to ten.

Note the asymmetry. Positive payment history accrues one month at a time; a single missed payment lands at full weight immediately and then decays over seven years. This is why the factor is worth 35% — it is the only input that is genuinely hard to fake.

The mechanism matters here. Most issuers do not report a payment as late until it is 30 days past due. Being three days late costs you a late fee and possibly a promotional rate, but usually not a credit-score entry. Being 31 days late costs you seven years.

Action: set up autopay for at least the minimum payment as a structural failsafe, then pay the full statement balance manually on top of it. Autopay for the minimum is not a repayment strategy — it is insurance against the 30-day cliff.

Amounts Owed — 30%

Usually discussed as credit utilisation: the percentage of your available revolving credit that you are using. If you have $10,000 in total limits and $3,000 reported in balances, utilisation is 30%. FICO's reasoning is straightforward — "if you are using a lot of your available credit, this may indicate that you are overextended."

The part almost nobody understands is when the number is captured. Your issuer reports a balance to the bureaux roughly once a month, and for most issuers that snapshot is taken on or near the statement closing date — not the payment due date. Pay in full every month and you can still report high utilisation, because the balance was measured before your payment landed.

Length of Credit History — 15%

This measures the average age of your accounts and the age of your oldest account. FICO notes that "in general, having a longer credit history is positive for your FICO Scores, but is not required for a good credit score" — which is the honest version of the advice, and the reason a thin file is a starting problem rather than a permanent one.

Your first card, opened years ago, is now your most valuable account from a history perspective. Closing it removes its age contribution and shrinks your total credit limit at the same time — a double hit, since the denominator in your utilisation ratio falls the moment the account closes.

Action: do not close old cards unless the annual fee is genuinely unrecoverable. Keep them alive with one small recurring charge and autopay.

New Credit — 10%

Every application creates a hard inquiry. Here the published detail is much less frightening than the folklore: FICO states that hard inquiries "stay on the report for up to two years, but they only affect the FICO Scores for a year," and that for most people "one additional credit inquiry will take less than five points off their FICO Scores."

The CFPB draws the hard/soft line clearly: soft inquiries — checking your own report, prescreened offers, existing-account reviews by your lender — do not affect your score at all. Only applications you initiate count.

Rate shopping is handled specially. FICO deduplicates multiple inquiries for the same loan type within a 14-day span on older score versions and a 45-day span on newer ones, and for mortgage, auto and student loans it "ignores inquiries made in the 30 days prior to scoring." Credit cards do not get this treatment — each card application stands alone. We cover the mechanics in more depth in what applying for a card actually does to your score.

Credit Mix — 10%

Lenders want evidence you can handle both revolving accounts (cards) and instalment accounts (loans with a fixed term). Cards supply only the revolving half. If your entire file is credit cards, the mix component is capped no matter how perfectly you behave.

This is the one factor where a purpose-built product genuinely helps. A credit-builder instalment account adds the missing tradeline without requiring you to borrow money you did not want. Do not open accounts purely to game this factor — at 10% it is the smallest lever on the board — but if you were going to take an instalment product anyway, the mix benefit is free.

Worked Example: Same Spending, Two Different Scores

Take two cards: Card A with a $4,000 limit, Card B with $6,000. Total available credit $10,000. You put $3,200 of ordinary spending through Card A this month and pay the statement in full, on time, every month. You pay no interest either way. Only the timing of your payment changes.

Identical spending. Identical cost. Identical discipline. A 26-point swing in the reported ratio, and the second row also avoids an 80% single-card figure, which several models weigh separately from the aggregate. The lever is a calendar reminder three days before your statement closes, not a change in how you live.

Worked Example: Why Revolving Destroys the Maths

The Federal Reserve's G.19 release put the commercial bank interest rate on credit card plans at 20.94% across all accounts in the second quarter of 2026, and 22.15% on accounts actually assessed interest. Use the second figure, because it describes people carrying a balance.

Carry $3,000 revolving for a year at 22.15%: $3,000 × 0.2215 = $664.50 in interest, roughly $55 a month.

Now the rewards side. Spend $1,500 a month on the same card — $18,000 a year — at a flat 2% cashback: $360. Net position: minus $304.50. You have run $18,000 through a rewards card, done everything the optimisation guides say, and finished the year worse off than if you had used cash.

This is the structural point. Card rewards are funded out of interchange and are worth low single-digit percentages. Card interest runs above 20%. The two are not the same order of magnitude, and no rewards strategy survives a carried balance. If you are already revolving, the score work waits — the balance comes first.

The grace period is what makes the difference, and it has conditions. The CFPB notes that issuers "are not required to give a grace period," that bills must be delivered at least 21 days before the due date, and — critically — that once you fail to pay in full, interest applies both to the unpaid balance and to new purchases "starting on the date each purchase is made." You do not just pay interest on the leftover. You lose interest-free status on everything until you clear the balance completely.

Who This Model Is Wrong For

The five-factor framework is FICO's, and FICO is a US product. It does not describe how you are assessed everywhere.

If you are outside the US. Most markets run credit reference agencies with different data and different rules, and lenders often apply their own internal scorecards on top. In the UK, for example, the binding consumer protection is not a score threshold but conduct regulation: the FCA's persistent debt rules require an issuer to intervene when a customer has paid more in interest and charges than principal over 18 months, escalating to a mandatory repayment arrangement or forbearance at 36 months. The regulator found roughly 4 million accounts in persistent debt, paying "around £2.50 in interest and charges for every £1 that they repay." Optimising a FICO number is meaningless there; avoiding that ratio is not.

If you have one card and a low limit. Utilisation maths is brutally volatile on a thin file. A single $900 purchase on a $1,000 limit reports 90%. The fix is not restraint, it is either a limit increase request (often a soft pull) or a mid-cycle payment — see building credit from scratch.

If the card does not report. Some cards — including several collateral-backed and crypto-secured products — issue without a hard credit check precisely because the lender's risk is covered by your collateral. No credit check often means no tradeline reported either, in which case the account builds nothing at all. Verify reporting before you rely on one of these for score-building, and treat them as spending tools rather than credit-history tools. The category is covered separately in our crypto card breakdown.

If you do not actually need credit. If you have no plan to borrow — no mortgage, no car finance, no business loan — a score above the point where you qualify for a decent card buys you nothing. Above roughly the prime threshold, further gains are largely cosmetic. Optimising past that is a hobby, not a financial strategy.

The Failure Modes

• Closing a paid-off card to "tidy up" — this shrinks your total limit and raises utilisation instantly
• Paying the statement in full but on the due date only, then wondering why utilisation still reports high
• Applying for three cards in a month because each individual inquiry "only costs a few points" — the compounding signal, not the points, is what underwriters react to
• Carrying a small balance deliberately to "show activity" — this is a myth that costs real interest and improves nothing
• Disputing an accurate late payment instead of building the 24 months of clean history that actually outweighs it

The Summary

Pay on time, every time — it is 35% of the model and the only input with a seven-year memory. Keep reported balances low by paying before the statement closes, not after. Keep old accounts open. Space applications out. Add an instalment tradeline if your file is cards-only, which a credit-builder product can do without new debt.

Applied consistently, that is enough. There is no shortcut, no repair service, and no sequence of applications that beats twenty-four months of clean reported behaviour — because a score is a prediction, and the only way to change a prediction is to change the data feeding it.