The verdict in three sentences
A single-provider checkout loses 8-14% of recoverable sales because a timeout or insufficient balance on provider A is never retried elsewhere. Failover means replaying the payment on provider B as soon as A returns a retryable error after 30 seconds. The real risk is not technical but financial: without an idempotency key, a mishandled retry creates a double charge that is costly in disputes.
Provider API SLA and availability
Before building failover, you need each API's reliability. These are 2026 orders of magnitude observed in Kenya.
| Provider | API uptime | Median latency | Recommended timeout | Average failure |
|---|---|---|---|---|
| M-Pesa | 99.8 % | 1.8 s | 30 s | 2-3 % |
| Airtel Money | 99.5 % | 3.2 s | 40 s | 3-4 % |
| Card (Visa/MC) | 99.2 % | 4.1 s | 45 s | 4-6 % |
| Bank transfer | 99.3 % | 3.8 s | 40 s | 4-5 % |
A 99.2% uptime means about 5.8 hours of downtime per month. In those windows, failover to a 99.8% provider saves the sale.
Decision table: retry or abandon
Not every failure should be replayed. Retrying an "insufficient funds" annoys the customer; retrying a "network timeout" saves them. Discipline lives in a code table.
| Error code | Meaning | Action | Retry on provider B |
|---|---|---|---|
| TIMEOUT | No response in 30 s | Retry | Yes |
| PROVIDER_UNAVAILABLE | Provider API down | Retry | Yes |
| INSUFFICIENT_FUNDS | Customer balance too low | Abandon | No |
| USER_CANCELLED | Customer declined the push | Abandon | No |
| INVALID_NUMBER | Wrong mobile money number | Abandon | No |
| DAILY_LIMIT_EXCEEDED | Daily cap reached | Offer other provider | Yes |
| PENDING_AUTH | Confirmation pending | Wait then verify | No |
Golden rule: only retry technical errors (timeout, unavailability, cap), never explicit customer declines.
Measured conversion gain
On a store handling 4,000 payment attempts per month, here is the impact of well-tuned failover. Average basket 15,000 FCFA-equivalent.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
| Metric | Without failover | With failover | Delta |
|---|---|---|---|
| Attempts/month | 4 000 | 4 000 | — |
| First-try success rate | 92 % | 92 % | — |
| Sales recovered by retry | 0 | 224 | +224 |
| Final success rate | 92 % | 97.6 % | +5.6 pts |
| Revenue recovered | 0 | 3 360 000 FCFA | +3.36 M FCFA |
Even at 5.6 points of recovery, the cumulative effect exceeds 3 million FCFA-equivalent in saved sales per month.
Mini case study
Wanjiru sells cosmetics online in Nairobi, 2,500 orders/month, average basket 18,000 FCFA-equiv. Before failover, 7% of her payments failed, half to timeout, i.e. about 87 lost sales (1,566,000). Adding an automatic M-Pesa-to-Airtel retry on timeouts, she recovers 60% of these technical sales: 52 orders and 936,000 per month. The idempotency rule prevents any double charge.
FAQ
How many retries before giving up? A single retry on a second provider suffices in 80% of recoverable cases. Beyond two attempts, you tire the customer and double-charge risk rises.
Does failover create double charges? Only if poorly designed. With a per-order idempotency key and a 24-hour dedup window, the risk drops to nearly zero.
Should I tell the customer about the switch? Yes, a clear message ("M-Pesa unavailable, trying card") reassures and cuts abandonment. Transparency lifts the retry's conversion rate.
What timeout per provider? 30s for M-Pesa, 40s for Airtel and transfer, 45s for card, matching their median latencies. Too short, you abandon wrongly; too long, the customer leaves.
Does failover work with cards? Yes, you can fall back from mobile money to a bank card as a last resort, useful for diaspora or large baskets above 100,000 FCFA-equiv.
Let's talk about your project. We design payment orchestration with M-Pesa, Airtel and card failover to recover your technical sales in Kenya. 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.
