Show me your scoring engine, I'll show you your loan book

Walk into any credit department at month-end, and the scene is familiar: a controller juggling between three phones, ageing reports piling up, and the quiet frustration of chasing money that left the business long ago.

We call it “collections.” We fund it, staff it, and measure it — yet we have the sequence wrong. After years of recovering debt for a living, I have learned this truth: by the time an account reaches my desk, the outcome was usually decided much earlier.

The money was not lost when the customer defaulted; it was lost when they were on boarded, when the scoring engine reviewed their Know Your Customer (KYC) and said “yes.” Everything after that is merely the bill arriving.

Put plainly, your receivables and loan book are shaped less by how hard you collect and more by the scoring engine you use to onboard. Get that gate right and recovery becomes easier. Get it wrong, and no collections army will save you.

For years, Kenyan lenders have relied on outdated scoring engines: an ID card, sometimes a payslip, a bureau check that mainly shows who has already failed, and an officer’s instinct. That engine was not built for the economy we serve today.

The trader managing M-Pesa float, the bodaboda rider, the Sacco member, and the SME waiting 90 days for a corporate invoice often have thin or invisible files.  A traditional engine either rejects the good customer you need or approves the bad risk you will meet eighteen months later in a lawyer’s suit. We call the result a collections failure. In truth, it is an onboarding failure that simply took time to mature.

What a real engine sees. A modern engine does not ask, “who has failed before?” It asks, “how does this customer handle money today?” It combines signals the old gate ignored: mobile-money patterns, till and POS turnover, utility-payment discipline, trade-credit behaviour, Sacco history, and telco data — all with the customer’s consent and through permissioned access, not surveillance.

It learns over time, flags applicants drifting toward distress before they miss a payment and identifies thin-file traders the bureau never saw. The effect compounds: a weak gate produces a weak book; a sharp gate quietly pays for itself again and again.

This is a boardroom decision, not an information technology assignment. A scoring engine is more than a software to deploy; it is the gate that either admits risk into the business or keeps it out before it damages cash flow, provisions, capital, and reputation.

It must therefore be governed with explainable decisions, clear audit trails, and bias-corrected rather than automated. An engine you cannot explain becomes an auditor’s concern; one you can explain becomes a competitive advantage.

I have said it often enough to risk repetition: African businesses rarely fail because opportunity is absent. They fail because liquidity is weak and credit systems are weaker. The graveyard is not filled with firms that lacked customers; it is filled with firms that admitted the wrong ones and exhausted themselves chasing payment.

So, ask one honest question at your next credit meeting: of the debt we are pursuing today, how much should never have been on boarded? Pull the file. The answer will discipline every decision that follows. Collections will always matter, but the strongest recovery strategy is not built after default; it is decided at the gate, before the first shilling leaves the door.

 The writer is a Certified Public Accountant and founder of Marathon Debt Recovery Ltd