Guide
How credit scores actually work
FICO and VantageScore, why they disagree, the five factors and what each one is really measuring.
Two families of score, and why they disagree
There is no such thing as “your credit score”. There are two major scoring families, several live versions within each, and three sets of underlying data. A lender picks one combination. You almost never find out which.
FICO is the older family and still the one most consumer lenders use. VantageScore was built by the three bureaus together and is the score most free consumer apps show you. Both run on a 300–850 range in their current versions, which is exactly why people assume they are the same number. They are not, and a 40-point gap between them on the same day is ordinary.
They disagree because they weight things differently and because they qualify differently. VantageScore can score a file with a single month of history; FICO generally wants at least six months of history and an account reported within the last six. If you are rebuilding, VantageScore will usually give you a number first, and it will usually be the more flattering one.
The practical consequence
The number in your banking app is real, but it is probably not the number a mortgage underwriter will pull. Use it to watch direction, not to predict an approval. If a lender quotes you a number that is 50 points below your app, neither of you is lying.
The five factors, and what each is actually measuring
FICO publishes approximate weights for the general population. They are not fixed per person — a thin file leans harder on length of history, and a recent serious delinquency swamps everything else — but they are the right mental model, and the two heaviest are the only two most people can move deliberately.
35% Payment history
Whether you paid on time, and how badly you did not.
The largest single input. A payment reported 30 days late costs more than one reported 29 days late costs nothing, because the furnisher only reports at 30. Recency and severity both matter: a 90-day late from four years ago weighs less than a 30-day late from last month.
Lever: Bring every open account current, then keep it current. Nothing else in this list outranks that.
30% Amounts owed
Mostly revolving utilisation — what you owe against your limits.
Utilisation is measured per card and across all cards, and it is calculated from the balance your issuer reports on the statement date, not the balance after you pay. This is why people who pay in full every month can still show 70% utilisation.
Lever: Pay before the statement closes, not just before the due date. This is the fastest lawful lever on the list.
15% Length of credit history
The age of your oldest account and the average age of all of them.
Closing an old card does not immediately delete its history — a closed account in good standing can keep reporting for around ten years — but it does remove its limit from your utilisation the moment it closes.
Lever: Keep the oldest card open and put one small recurring charge on it. Time is the only other input.
10% New credit
Recent hard inquiries and recently opened accounts.
A hard inquiry stays on the report for two years but is generally only scored for one. Rate shopping for a mortgage or a car within a short window is usually treated as a single inquiry by the scoring model; applying for four credit cards in a month is not.
Lever: Space out applications. Do not open new accounts in the ninety days before a mortgage application.
10% Credit mix
Whether you have both revolving and instalment accounts.
The smallest factor and the one most oversold. It is not worth taking on a loan you do not need in order to improve it.
Lever: Effectively none worth pulling on its own. It improves as a by-product of normal borrowing.
Weights are FICO’s published approximations for the general population. Your own file may weight them differently — someone with a thin file leans harder on length of history, and a recent serious delinquency dominates everything else.
Factor weights, as a table
| Factor | Weight | What it measures |
|---|---|---|
| Payment history | 35% | Whether you paid on time, and how badly you did not. |
| Amounts owed | 30% | Mostly revolving utilisation — what you owe against your limits. |
| Length of credit history | 15% | The age of your oldest account and the average age of all of them. |
| New credit | 10% | Recent hard inquiries and recently opened accounts. |
| Credit mix | 10% | Whether you have both revolving and instalment accounts. |
What people get wrong about each one
Payment history is binary at the 30-day line. A payment three weeks late is not reported at all; a payment 30 days late is. If you are going to be late, being late by 29 days is materially different from being late by 30, and it is worth a phone call to the creditor to find out which side of the line you are on.
Amounts owed is not “how much debt you have”. It is dominated by revolving utilisation, computed from the balance your issuer reports — normally the statement balance. This is why people who pay in full every month can still show 70% utilisation: they pay after the statement closes, so the high figure is the one that reaches the bureau. It is the fastest lawful lever on the list and the one most people have never been told about.
Length of credit history is the age of your oldest account and the average age of all of them. Opening a new account lowers the average; that is the real cost of a new card, not the inquiry. Closing an old account does not immediately delete its history — a closed account in good standing can keep reporting for around ten years — but it removes its limit from utilisation the same day.
New credit is a small factor that gets outsized attention. A hard inquiry stays on the report for two years but is generally only scored for about one, and typically costs a handful of points. Rate-shopping for a mortgage or a car within a short window is usually treated by the model as one inquiry; applying for four credit cards in a month is not.
Credit mix is the smallest and the most oversold. It improves as a by-product of ordinary borrowing. Nobody should take on a loan they do not need in order to move a 10% factor.
Three files, three numbers, identical behaviour
Furnishers choose which bureaus they report to, and many report to one or two rather than all three. So the same person, behaving identically, produces three different files and three different scores. The chart below is a demonstration file: the gap between the highest and lowest bureau moves between 8 and 23 points across a year, with nothing changing in what the person actually did.
Three bureaus, one sample file, twelve months
Every value, as a table
| Point | Experian | Equifax | TransUnion |
|---|---|---|---|
| Sep | 612 | 598 | 621 |
| Oct | 610 | 601 | 619 |
| Nov | 615 | 599 | 624 |
| Dec | 618 | 606 | 622 |
| Jan | 617 | 611 | 628 |
| Feb | 624 | 610 | 627 |
| Mar | 629 | 618 | 626 |
| Apr | 631 | 622 | 633 |
| May | 636 | 620 | 638 |
| Jun | 634 | 627 | 641 |
| Jul | 641 | 631 | 639 |
| Aug | 644 | 629 | 646 |
This matters when you apply for something. Mortgage lenders typically pull all three and use the middle score. Auto lenders often use one bureau, and which one can depend on the region. Card issuers vary by product. If one of your three files is materially worse than the others, the honest first question is not “how do I raise my score” but “which of the three is short of information, and why”.
Which version a lender is actually looking at
Within FICO alone, a mortgage lender may still be pulling the classic versions mandated for conforming loans, an auto lender an industry-specific auto score on a 250–900 range, and a card issuer FICO 8 or FICO 9. The differences are not cosmetic: FICO 9, for instance, ignores paid collections and weights medical collections less heavily, and FICO 10 T looks at a trended 24-month view of your balances rather than a single snapshot.
None of this is something you can control, and none of it is something to optimise for. It is the reason to be sceptical of any tool — ours included — that hands you a single confident number.
What any of this means for you
Three things, in order of size:
- Bring every open account current and keep it current. Nothing else on the list outranks 35%.
- Find out when each revolving account reports, and get the balance down before that date rather than before the due date. This is free, immediate, and reversible.
- Stop opening things you do not need, particularly in the ninety days before a mortgage application.
Everything else — disputes included — is either a correction of something that is wrong or a waiting game. Disputes are worth running when the file contains errors. They are not a score strategy, and a company that sells them as one is selling you the wrong thing.