The verdict in three sentences
Mobile money cuts chargebacks but not fraud: fake orders, promo abuse and account takeover run at 1.5–4% of orders in Africa. A rules engine — hard blocks plus a review queue with scoring — filters risk before payout, where manual checking can't scale. Blocking even 1% of fraud on 20,000,000 FCFA/month saves 200,000 FCFA outright, at a manageable false-positive cost.
Hard block vs review queue
Two logics coexist. Hard-block rules immediately reject a certain signal (blacklisted number, >5 CVV failures in 10 min). Review scoring assigns a 0–100 score and queues ambiguous orders (review above 60, block above 85). The combination limits false positives while stopping obvious fraud.
| Signal | Rule | Action | Expected false positives |
|---|---|---|---|
| >3 orders/hour/number | Hard velocity | Temporary block | Low (~2%) |
| >5 CVV failures/10 min | Hard block | Reject + cooldown | Very low (~1%) |
| Mobile money name ≠ delivery name | Score +30 | Review queue | Medium (~10%) |
| Same device, 4+ accounts | Score +40 | Review queue | Medium (~8%) |
| Promo code reused (multi-account) | Score +25 | Promo block | Low (~5%) |
| Known risky address | Score +35 | Manual review | Medium (~12%) |
Velocity and 0–100 scoring
Scoring adds up signals and triggers the action per threshold. The goal: automate 90% of decisions and leave only the grey zone to humans.
| Cumulative score | Interpretation | Automatic action | Share of volume |
|---|---|---|---|
| 0–40 | Low risk | Direct approval | ~85% |
| 41–60 | Watch | Payout delayed 24h | ~8% |
| 61–85 | Suspicious | Manual review queue | ~5% |
| 86–100 | Likely fraud | Automatic block | ~2% |
*2026 orders of magnitude; COD adds 12–25% refusal/no-show to monitor.*
Mini case study
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Wanjiku runs a mobile-money store in Nairobi taking 20,000,000 FCFA/month (KES equivalent) with an estimated 2% fraud rate, i.e. 400,000 FCFA of potential losses. Installing velocity rules (max 3 orders/hour/number) and scoring with a review threshold above 60, she blocks half the fraud, saving 200,000 FCFA/month. The cost: a few minutes of manual review on ~5% of orders and ~2% false positives recovered via a WhatsApp contact — a comfortably profitable ratio.
FAQ
What's the e-commerce fraud rate in Africa? In 2026 it sits between 1.5% and 4% of orders by sector, with COD fraud (refusal/no-show) adding 12–25% of unfulfilled orders.
Which velocity rules come first? Cap at 3 orders/hour/number and block beyond 5 CVV failures in 10 minutes: these two rules stop most bots and card testing.
What scoring threshold should I use? A 0–100 score with review above 60 and automatic block above 85 automates ~90% of decisions while keeping a human eye on the grey zone.
How much does blocking fraud earn? Blocking 1% of fraud on 20,000,000 FCFA/month saves 200,000 FCFA outright; over a year that's 2,400,000 FCFA preserved for a modest setup cost.
How do I handle false positives? Use a 24h delayed payout for scores 41–60 and a WhatsApp verification contact: you recover the legitimate sale without letting fraud through.
Become a Kolonell referral partner
Know merchants losing money to fraud and no-shows? Introduce them to Kolonell and earn 12% on an e-commerce build, 15% + 5% recurring on a showcase site, 10% on a marketplace and 8% on institutional. The referral (apporteur d'affaires) programme pays for every successful introduction.
Let's talk about your project. We install your anti-fraud engine to stop fake orders before payout. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
