The verdict in three sentences
On Black Friday it isn't traffic that's missing: it's the checkout that buckles and turns ready-to-pay customers into abandoned carts. With an 8x to 15x surge over 48 hours and USSD failure rates jumping from 8 % to 20-30 %, every hour of checkout downtime costs 5 to 15 % of daily revenue. The fix is not betting everything on one PSP but building a chain of queue + retry + backup provider.
One PSP or a failover?
Betting everything on a single provider means its outage becomes yours. Failover automatically switches to a second provider (M-Pesa→Airtel Money, Paystack→Flutterwave) when the first crosses a failure threshold, with the customer none the wiser.
| Strategy | Resilience | Complexity | Peak loss |
|---|---|---|---|
| Single PSP | Low | Simple | High |
| PSP + idempotent retry | Medium | Moderate | Medium |
| PSP + secondary failover | High | High | Low |
| Multi-PSP + queue | Very high | High | Very low |
The costed 2026 peak playbook
Goal: keep under 2 % application-level failure at peak. Here are the levers and their targets.
| Lever | 2026 target |
|---|---|
| Pre-event load test | 3x expected peak |
| Checkout uptime | 99.9 % |
| Webhook | Idempotent (0 double charge) |
| Automatic retry | 2-3 spaced attempts |
| Provider failover | Switch < 5 s |
| Target app failure rate | < 2 % |
| Cache + rate limiting | Active on product pages |
Under load, gateway timeouts and USSD saturation are the top two failure causes. Idempotent retry prevents double charges, and caching absorbs the read spike on product pages.
Mini case study
Wanjiru, who runs a fashion store in Nairobi, does KES 900,000 in revenue on Black Friday day. With no failover, her single PSP goes down for 2 hours at peak: at 12 % of daily revenue per hour, she loses ~KES 216,000. After adding a backup provider and idempotent retry (build cost ~KES 130,000), downtime falls to a few minutes and the loss to under KES 18,000: the investment pays off on the very first Black Friday.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
FAQ
Why does the USSD failure rate climb so much at peak?
The USSD channel shares limited capacity across all merchants. At Black Friday peak the queue saturates and failure jumps from ~8 % to 20-30 %. A backup provider and retry reduce the customer-side impact.
What is an idempotent webhook and why is it critical?
An idempotent webhook processes the same payment event only once, even if received multiple times. Without it, a retry can charge the customer twice — unacceptable at peak when retries multiply.
At what level should I load-test before a sale?
Aim for 3x expected peak. If you expect a 10x traffic surge, test at 30x: it exposes gateway timeouts and bottlenecks before they cost sales on the day.
How much does an hour of checkout downtime really cost?
Between 5 and 15 % of daily revenue depending on timing. During a 48-hour peak, the busiest hour alone can concentrate more than 10 % of the day's sales.
Let's talk about your project. We audit your checkout and set up failover, idempotent retry and load testing before your sale. 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.
