A 40-point gap between the score in your banking app and the score your lender quoted is normal. Neither is wrong. They are different measurements.
Three sources of difference, all at once
Different bureaux. Experian, Equifax and TransUnion each hold their own file. Lenders are not obliged to report to all three, and many report to two. An account that appears on one file and not another changes utilisation, account age and account count — and therefore the score.
Different models. FICO and VantageScore are separate companies with separate formulas. Both output a number from 300 to 850, which makes them look comparable, and they are not.
Different versions. FICO 8 is the general-purpose standard. FICO 9 and 10 exist. And mortgage lending, by regulation, still uses FICO 2, 4 and 5 — models from the late 1990s and early 2000s. Those older models treat collections, authorised-user accounts and medical debt differently from FICO 8, and typically produce a lower number.
Three bureaux times several models times several versions is why "your credit score" is not a single fact.
Which number is used when
| Purpose | What the lender usually pulls |
|---|---|
| Mortgage | FICO 2, 4 and 5 from all three bureaux — the middle score of the three is used |
| Car finance | FICO Auto Score, an industry version that weighs car-loan history more heavily |
| Credit cards | FICO Bankcard Score, or FICO 8 |
| Renting | Varies widely; often VantageScore, sometimes no score at all |
| Free banking apps | Usually VantageScore 3.0 or FICO 8, from one bureau |
The mortgage case is the one that catches people. Three bureaux are pulled, the middle score is taken, and if two borrowers apply jointly the lower of their two middle scores usually governs the rate.
What actually moves the number
The weightings differ between models, but the shape is consistent:
- Payment history, roughly 35%
- Amounts owed, chiefly utilisation, roughly 30%
- Length of credit history, roughly 15%
- Credit mix, roughly 10%
- New credit, roughly 10%
The first two are about two thirds of the total. Everything else is noise by comparison, which is worth remembering before optimising credit mix.
Utilisation is the fast lever
Utilisation is the share of your available limits currently in use. It recalculates every time a balance is reported — typically on the statement date, not the due date.
That timing is the whole trick. Paying a card down before the statement closes reports a lower balance. Paying it in full after the statement closes reports the higher one, even though you never carried debt and never paid interest.
Someone who pays their card in full every month can still show 60% utilisation if they spend heavily and pay after the statement. Moving that payment earlier can move a score within one cycle, with no change in behaviour.
Payment history is the slow one
A missed payment reported to the bureaux stays on file for seven years. Its weight fades well before then — a late payment from five years ago counts for far less than one from five months ago — but it does not disappear on request.
Which is why a single 30-day late is worth avoiding at some cost. Most lenders do not report until an account is 30 days past due, so a payment a week late is usually invisible to the bureaux even if it attracts a fee.
Closing an old card usually hurts
It removes that limit from the utilisation calculation, which raises utilisation on everything else immediately. Over time it also shortens the average age of accounts.
If an annual fee is the reason, ask the issuer to downgrade the card to a no-fee product in the same family. The account, its limit and its history survive.
Checking your own score changes nothing
A soft enquiry — checking your own report, most prequalification tools, most rental applications, and an issuer reviewing an existing account — has no effect on the score.
A hard enquiry, from an actual credit application, typically costs a few points and fades within a year. Rate shopping is deliberately protected: multiple mortgage or auto enquiries within a short window, 14 days in older FICO models and 45 in newer ones, count as one. Comparing five lenders in a fortnight is close to free; spreading the same five over four months is not.
The number is not the decision
Lenders do not approve scores, they approve applications. Debt-to-income ratio, employment history, deposit size, and the asset itself all sit alongside the score, and a strong file elsewhere is regularly approved at a number that would fail on its own.
The score decides the price more often than it decides the answer.