Larger companies have teams for this. Companies of thirty do not, so the losses show up as a bad month that nobody quite explains. The good news is that most of what a risk team does first is available to you cheaply.

Start with rules, and measure them

A short list of well-chosen rules catches a surprising share of the problem: the new account ordering far above the norm, the delivery address that does not match anything, the customer whose payments have quietly slipped from seven days late to forty. We write them, run them against your last two years, and show you what they would have caught and what they would have wrongly stopped.

That second number matters more than the first. A rule that blocks fraud and also blocks your best customer costs you money in a way that is harder to see.

Then a model, if the rules leave enough on the table

When you have enough history and the patterns are genuinely tangled, a model adds real value on top. We only propose one after the rules are running, so there is something honest to compare it to.

Credit and collections

For anyone selling on terms, the bigger loss is usually not fraud — it is slow payment and eventual write-off. Both are visible in payment behaviour months ahead. A weekly list of accounts that are drifting, ranked by exposure, is worth more than any single model, and takes weeks not quarters.

Where we stop

Every rule and every score is reviewable, and a person makes the decision to decline, hold or chase. We do not build systems that silently refuse a customer, partly because it is wrong and partly because you will eventually have to explain one to somebody.

Send us two years of orders and payments

We will show you what a rule set would have caught.