The verdict in three sentences
Fraud is not inevitable: 80% of attempts come from bots and scripts that a single velocity rule stops cold. A merchant who stacks three layers — automated rules, 3D Secure on cards and device verification — brings losses below 0.3% of revenue. Mobile money and cards do not share the same risk profile, and treating both with identical rules costs you either sales or money.
Fraud signals and automatic action
A good anti-fraud policy does not say "yes" or "no": it assigns a score and triggers a graduated action. Here are the most profitable signals to watch in 2026, with the alert threshold and recommended response.
| Fraud signal | Alert threshold | Automatic action |
|---|---|---|
| Velocity (transactions/card/hour) | > 3 attempts | Temporary 60-min block |
| Abnormal amount vs average basket | > 5x average basket | Manual review + forced 3DS |
| Multiple cards, same device | > 2 cards / 24h | Device block |
| IP / country mismatch | IP ≠ card country | 3DS challenge |
| Momo number on blacklist | Exact match | Immediate decline |
| Repeated failures then success | > 4 failures | Risk score +40 |
| E-mail created < 24h | New account | First-order cap |
The principle: a single signal triggers a challenge (3DS, OTP), several stacked signals decline. This avoids blocking a loyal traveling customer while stopping the fraudster testing 20 stolen cards.
Mobile money vs cards: two risk profiles
In Lagos as in Dakar, card fraud and mobile money fraud look nothing alike. Cards suffer data theft and chargebacks; mobile money suffers social engineering (fake SMS, refund scams) but almost never chargebacks.
| Criterion | Card fraud | Mobile money fraud |
|---|---|---|
| Main vector | Stolen data, testing | Social engineering, fake agent |
| Chargeback possible | Yes | Rare (proof-based dispute) |
| Average incident cost | 25,000 - 80,000 FCFA | 10,000 - 40,000 FCFA |
| Dispute window | 7 to 45 days | 24h to 7 days |
| Key defense tool | 3D Secure 2 | OTP + device verification |
| Share of attempts (2026 ballpark) | 55% | 45% |
Mini case study
Chinedu runs an online electronics store in Lagos with 12,000,000 FCFA in monthly revenue. Before protection he lost 2% to fraud, or 240,000 FCFA/month. After turning on the velocity rule, 3DS on baskets over 100,000 FCFA and a number blacklist, his fraud rate dropped to 0.3%, or 36,000 FCFA/month. Net saving: 204,000 FCFA/month, nearly 2,448,000 FCFA/year, for a one-time setup cost far below that.
FAQ
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Does 3D Secure scare customers away?
Misconfigured, yes: it can add 10-20% cart abandonment. Well configured with a low-value exemption, it only challenges risky transactions and recovers more sales than it loses.
How long to set up anti-fraud rules?
On a modern store connected to a PSP, plan 2 to 5 days of configuration. Velocity and blacklist rules are often included on the PSP side; the rest is setup.
Is a small store really targeted?
Yes. Bots don't sort by size: they look for unprotected checkouts to test stolen cards. A store with 2,000,000 FCFA revenue can lose 30,000 to 60,000 FCFA/month unknowingly.
Do I need a dedicated manual-review team?
No, not at first. A "manual review" queue of 2 to 5 orders/day takes 15 minutes. Only beyond several dozen suspicious orders/day does a dedicated role make sense.
Since mobile money has no chargebacks, can I ignore fraud?
No. You avoid chargebacks, but social engineering and fake refunds cost 10,000 to 40,000 FCFA per incident. Device verification and OTP remain essential.
Let's talk about your project. We'll configure your anti-fraud rules and 3DS to block bots without breaking sales. 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.
