The cost of outdated credit scores

A goalkeeper who saves every shot looks brilliant until you realise the team never scores. That's what's happening in trade credit. The entire system is built around preventing the worst outcome, and nobody's counting the goals that are never taken.

When suppliers think about bad credit decisions, they think about bad debt. A customer who didn't pay, an invoice written off, a hit to the bottom line. But bad debt is only half the equation. A bad credit decision cuts both ways, and the side nobody measures is almost always the more costly one.

Two ways to get it wrong

Every time a credit team evaluates a trade account application, there are two possible mistakes.

The first is saying yes to the wrong customer. This is the cost suppliers know intimately. In UK construction, more than 70% of firms carry bad debt, with the average supplier absorbing roughly £10,000 a year in write-offs. Industry-wide, that totals around £2 billion annually. And when a major contractor collapses as ISG did in 2024, leaving over £700 million owed to suppliers and subcontractors — the fallout ripples across entire supply chains.

The second is saying no to the right customer, or dragging the process out so long that they leave before a decision ever lands. This cost never hits the balance sheet. There's no write-off for "revenue we didn't earn because we turned away a perfectly good buyer." But it's real, it's large, and it compounds quietly every single month.

Putting a number on the invisible cost

Think of it like an iceberg. Bad debt is the part above the waterline, visible, tracked, reported on. The lost revenue from over-cautious decisions sits below the surface, unseen and unaccounted for. And like any good iceberg, the part you can't see is far bigger than the part you can.

A construction buyer placing £4,000 in orders per month represents £48,000 in annual revenue. If that buyer stays for five years, which is perfectly normal in a well-managed trade relationship, their lifetime value reaches £240,000.

When that buyer gets wrongly declined, or the approval drags on long enough that they take their business elsewhere, the supplier doesn't lose a single order. They lose a quarter of a million pounds in future revenue. And they'll never know, because the customer simply never appears in their system.

Now scale it up. A supplier processing 50 new credit applications a month only needs to lose 10% of viable applicants, just 5 a month, to bleed 60 good customers a year. At £240,000 in lifetime value each, that's £14.4 million in revenue that evaporated before it ever began.

Set that against the £10,000 in average annual bad debt. The invisible cost of turning away good customers dwarfs the visible cost of approving bad ones.

Why the imbalance persists

If false declines are so much more expensive than bad debt, why do suppliers keep optimising for the wrong side?

Because bad debt is loud. When a customer defaults on a £30,000 invoice, the whole business knows about it. There's a write-off. There's a difficult conversation with the finance director. There's a post-mortem and a tightening of the rules.

A lost customer makes no sound at all. Nobody calls a meeting to discuss the buyer who applied for a trade account on Tuesday and placed an order with a competitor on Thursday because the approval hadn't come through. That buyer never appears in a single report, dashboard, or quarterly review. In the supplier's world, they simply don't exist.

This is what makes the problem so persistent. Credit teams are measured on what they can see, bad debt ratios, days sales outstanding, write-off percentages. Nobody measures what they can't — the customers who never materialised, the orders that were never placed, the relationships that never had a chance to form.

So the system drifts steadily in one direction. Every bad debt tightens the criteria. Every insolvency headline makes the team a little more cautious. Meanwhile, the pile of missed opportunities on the other side of the ledger keeps growing, completely invisible.

The 93/7 problem

Here's a number that should reframe the entire conversation: 93% of trade credit extended in the UK gets repaid. The whole apparatus, credit reports, bureau scores, manual checks, multi-day approval processes is engineered around the 7% that doesn't.

That 7% matters. No one is suggesting otherwise. But when a system designed to contain the 7% actively constrains growth from the 93%, the economics fall apart. The cost of over-protection ends up exceeding the cost of what it's protecting against.

A credit team that declines one good customer in order to avoid one bad one hasn't broken even. They've traded five years of revenue from the good customer to dodge a single write-off from the bad one. That's not risk management. That's value destruction in disguise.

What a balanced approach looks like

The answer isn't to lower the bar and accept more defaults. It's to build a complete picture of every customer, one that draws on live banking data, real-time payment behaviour, and operational signals, not just a credit score filed 18 months ago. Modern machine learning models can read these signals continuously, spotting trajectory and momentum rather than relying on a rear-view mirror. When the credit team can see the full story as it unfolds, not a static summary of where a business used to be, they stop guessing and start making decisions grounded in what's actually happening right now.

In practice, that comes down to three shifts.

The full story, not a fragment of it. Traditional credit reports pull from a single source at a single point in time, a Companies House filing from 18 months ago, a credit score that compresses an entire business into one number. The full credit story looks nothing like that. It's live banking data showing real cash flow. Its payment behaviour tracked across suppliers over months. Its operational signals, new contracts won, directors changed, invoices accelerating or slowing. When a credit team can see all of this in one place, continuously updated, they're not making a leap of faith. They're reading a story that's already been written.

Foresight, not hindsight. Most credit tools tell you where a business has been. Modern machine learning models can tell you where it's heading, recognising patterns in payment trends, revenue trajectory, and behavioural signals that a human reviewing static data would never catch in time. A customer whose payments are slowing by a few days each month looks fine on a quarterly review. To an ML model tracking the pattern in real time, it's an early warning. The same works in reverse, a buyer whose behaviour is steadily improving is a growth opportunity that static reports will never surface.

Measuring both sides of the ledger. This is the cultural shift that makes everything else stick. Credit teams have always been measured on what they prevented, bad debt ratios, write-off percentages, days sales outstanding. But if nobody tracks what was enabled, approval rates, time-to-decision, the percentage of approved customers who become repeat buyers, then the invisible cost stays invisible. The moment a business starts measuring growth unlocked alongside risk avoided, the credit function transforms from a defensive line into a revenue driver.

The decision that really matters

Every credit decision is, at its core, a bet on a relationship. Say yes to the right customer and you're backing years of growing orders, reliable payments, and compounding value. Say no, and you're wagering that the £10,000 you might save in bad debt is worth more than the £240,000 in lifetime revenue you'll never see.

When the data is stale and the process is manual, that bet feels like a coin toss. When the intelligence is current, continuous, and complete, it starts to look a lot more like an informed decision.

The true cost of a bad credit call isn't just the invoice that goes unpaid. It's the customer who never arrives, the orders that are never placed, and the growth that never happens. The suppliers who learn to see both sides of the equation are building a credit process that optimises for growth.