Guide for risk teams, credit managers, CFOs
Company's ability to pay debts long-term. Analyzed via balance sheets, financial ratios, and accounting deterioration signals.
Dependency on a few key accounts. Customers > 20% of revenue are highly concentrated. Risk of sudden cash flow drop.
Payment history, disputes, legal proceedings. A good balance sheet can hide a chronic bad payer.
Editorial Note
The most underestimated risk on SMEs is behavioral: a solvent company on paper can be a structural bad payer.
What data? Where to find it (SIREN, public balance sheets, BODACC, banking data via open banking).
Calculate risk level by combining financial ratios and weak signals (supplier delays, director rotation, legal proceedings).
APPROVED / REVIEW / DECLINED grid + credit limit and payment terms adapted to risk.
Configurable email alerts if risk indicators deteriorate. Re-evaluation at least once a year.
💡 Good to know
Behavioral signals (late payments, disputes, falling collections) often appear before the balance sheet deteriorates.
Result:
0-2 yes → Insufficient data — REVIEW risk
3-4 yes → Partial analysis possible — Request supplements
5-6 yes → Complete analysis — Reliable decision
< 10 files/month
Manual analysis sufficient
10-50 files/month
Semi-automation recommended (simple scoring + human validation)
> 50 files/month
Automation necessary (API scoring)
⚠️ Regulatory Alert
Beyond efficiency, AI Act requires any high-risk automated assessment system to be explainable and auditable. Automation without compliance is a legal risk.
Myth 1: "Good balance sheet = good customer"
Myth 2: "Scoring is enough to decide"
Reality: Borderline cases (scores 40-60%) always require human supervision and supplementary qualitative analysis.
Myth 3: "Initial assessment is enough"
Reality: a customer who was solvent at onboarding can deteriorate later. Regular follow-up is essential, not optional.