How Credit Scores Actually Work: The Five Inputs and What Moves Them Most

A credit score is a three-digit number that decides things far bigger than three digits should: whether a landlord returns your call, what interest rate a lender quotes, sometimes even your car insurance premium. And yet most people learn how it works the way you learn rules to a board game you've already lost — one penalty at a time.

Here's the whole machine, up front: a credit score is a prediction. It uses the information in your credit report to estimate how likely you are to fall seriously behind on a payment in the next couple of years. That's it. It doesn't know your salary, your savings, or your character. It knows how you've handled borrowed money, and it weighs five specific things.

The five inputs, weighted

The most widely used scoring formula in the US is FICO, which runs from 300 to 850; the Consumer Financial Protection Bureau explains the model's role in lending decisions here. FICO publishes the recipe's proportions on its education site:

1. Payment history (35%): the only factor that's a dealbreaker

The formula's biggest question is the simplest: have you paid your bills on time? On-time payments, month after month, are the slow engine of a good score. The reverse is sharply asymmetric — a payment reported 30 or more days late can undo years of quiet on-time history, and the damage grows at 60 and 90 days. (Being a few days late usually triggers a fee but doesn't reach your credit report; the full sequence is worth understanding, and we've mapped the entire 30/60/90-day timeline here.)

Think of payment history like a driving record: a decade of clean driving builds slowly, one crash shows up instantly, and the record forgets on a schedule — most negative marks fall off your report after seven years under federal law.

2. Amounts owed (30%): how full your tank looks

The second factor is mostly about your credit utilization ratio — your card balances divided by your card limits.

Worked example. Jordan has two cards: one carrying $500 against a $1,000 limit, another carrying $300 against a $2,000 limit. Total: $800 of $3,000, a 27% overall utilization — though that first card alone sits at 50%, and scoring models look at both the total and each card. Now watch two moves:

• Jordan pays the first card down to $100. New total: $400 of $3,000 — 13%. Utilization improves within a month or so, because card issuers typically report balances monthly. This is the fastest-moving lever in the entire score.
• Instead, Jordan closes the unused older card after paying it off. Limits drop from $3,000 to $1,000, and the same $500 balance jumps from 17% to 50% utilization. Nothing about Jordan's habits changed; the ratio's denominator did.

One crucial decoupling: utilization is measured from the balance your issuer reports, not from whether you carry debt month to month. You can pay in full every cycle — never owing a cent of interest — and still show healthy utilization. Interest is a cost; utilization is a snapshot.

3. Length of credit history (15%): the factor you can't rush

This measures how old your accounts are — oldest, newest, and the average. It's why the first card someone opens matters years later as the anchor of their file, and why closing your oldest account can eventually shorten your visible history. There is no trick here, only time; a starter card like a secured card exists mostly to start this clock.

4. New credit (10%): applications leave footprints

When a lender pulls your report because you applied for credit, that's a hard inquiry, and it can trim a few points for a while. Checking your own score, or a background check, is a soft inquiry — invisible to the formula. Scoring models also treat rate shopping sensibly: multiple auto loan or mortgage inquiries within a short window are typically counted as one search, because comparing rates on one car loan isn't the same behavior as opening five credit cards in a month.

5. Credit mix (10%): a small preference for range

The formula gives modest credit for handling different kinds of accounts — revolving credit (cards) and installment loans (auto, student, mortgage). It's the least important factor and never worth borrowing money just to demonstrate. It mostly explains why someone's score ticks up over the years as ordinary life adds an installment loan to a card history.

Myth vs. fact, rapid fire

What the number gets you: reading the bands

Lenders don't experience your score as a precise number; they experience it as a tier. The industry's rough bands for FICO's 300–850 scale run: below about 580 "poor," 580–669 "fair," 670–739 "good," 740–799 "very good," 800+ "exceptional." The tiers are where the money is: in Experian's Q1 2026 auto-lending data, a prime borrower (661–780) averaged 8.77% APR on a used car loan while a near-prime borrower (601–660) averaged 14.03% — on a typical $27,000 loan over six years, that's a difference of more than $5,000 in interest for a few dozen points of score. The score isn't a grade on your character; it's a price tag on your borrowing.

Two more mechanics worth knowing about the number itself. First, you have more than one: each bureau's file can differ (lenders don't all report to all three), and FICO alone maintains multiple versions, plus industry-specific variants for auto lenders and card issuers. A 12-point gap between two apps' scores is normal, not an error. Second, starting from nothing is its own state: with no accounts at all, you don't have a low score — you have no score, because there's nothing to predict from. The formula generally needs an account that's been open and reported for several months before it can produce a number, which is why first credit products matter out of proportion to their size.

What moves the needle most, honestly ranked

1. Never letting an account go 30 days past due. Nothing else you do matters as much, in either direction.
2. Reported utilization. The one factor that can improve within a single billing cycle.
3. Time. Old accounts quietly aging are doing more work than any hack.
4. Restraint with applications. A few points each, briefly — meaningful mostly right before a major loan.
5. Mix. Barely. Let it happen naturally.

The quiet takeaway: a credit score isn't a judgment of you. It's a narrow, mechanical prediction built from five inputs, three of which you can influence this year and one of which you can influence this month. Once you know the weights, the mysterious number becomes what it always was — arithmetic.