The fraud hiding in your first payment defaults

A default with no repayment is rarely a scorecard failure. It is first party fraud, booked as credit loss because nothing at the point of application explained it.

Tuesday, Wednesday, Thursday

1. Tuesday: someone applies for a credit card.
2. Wednesday: they apply again, through a different lender, now with two dependants instead of three.
3. Thursday: it is one dependant, a job title upgrade, and income a few hundred pounds higher.

Each lender sees one clean application. The details are plausible, the file checks out, and somebody lends. Weeks later that loan becomes a first payment default, and a credit team starts asking what their model missed.

The model missed nothing. Each application, viewed alone, was fine. The fraud was in the sequence, and the sequence was invisible, because no single lender can see the applications made to everyone else.

No repayment is fraud

A credit risk lead at a UK consumer lender gave us the cleanest working definition we have heard:

"No repayment is fraud. If the customer made a repayment, then it's credit."

Most lenders book all of it as credit loss. That is not an accounting quirk. It means the fraud has no owner, no budget line and no fix. The number sits in provisions, quietly compounding, while the fraud tooling conversation stalls because officially there is no fraud problem.

Intent shows before the first payment is missed

The Tuesday applicant is not an edge case. When someone never intends to pay, the intent leaks into the data before they reach you:

• Details that shift between applications.
• Amounts that escalate.
• A story that changes depending on who it is being told to.

None of it appears in a bureau file.

Infact screens every application in real time, across every lender reporting into our bureau. Each application resolves to a single verified identity, matched even as names, addresses and contact details shift. One person, one view, however the story is told. The detection focus is data manipulation, and the distinction matters: shopping around for credit over time is healthy, considered behaviour, and the data treats it as a positive signal. Manipulating your story between applications is intent.

Consortium fraud databases match applications against fraud after it has been confirmed. Intent is readable while the application is still in front of you.

What one lender found

A UK personal loans lender asked us to screen their historic applications against the Infact bureau and their own confirmed fraud outcomes. Their team validated the results.

The combined signals caught around 50% of their confirmed paid frauds. The exercise identified roughly £500,000 of fraud, against their own modelled cost of about £9,000 per accepted fraudulent customer. Since the exercise, detection has run ahead of forecast as more frauds have been confirmed. They are now moving to a live shadow deployment.

The economics of protection are broken

Fraud leaders tell us the same thing about the established tools: they are priced for banks. Six-figure price tags for incumbent solutions that nobody can build a business case around. Monthly fees that only make sense after the attack has happened. So growing lenders run thin screening and absorb the losses into the credit line, where nobody has to defend them.

When an accepted fraudster costs you thousands and prevention costs pennies per application, the trade‑off should be obvious. Yet many tools are priced so the cost of preventing fraud can approach (or even exceed) the cost of absorbing it.

That pricing gap is the problem, it shouldn’t be where the value ends up.

Prove it on your own book

We prove the detection rate before any commercial conversation. Screen your historic applications against your confirmed fraud outcomes, see exactly what would have been caught and what would have been referred, and measure the false positive cost of every detection strategy. Where a signal only proves itself live, we run in shadow mode and measure against outcomes as they mature.

If you have a first payment default number nobody can explain, that is where we would start.