How to evaluate an African payments API
Coverage maps all look the same: forty countries, every method, five nines. The evaluation that predicts what production will feel like asks six questions: how coverage is counted, whether delivery numbers are measured or promised, what the funding model traps, whether collections and payouts share one balance, how deep local methods really go, and who regulates the counterparty. This guide gives you the six questions and what a good answer looks like, so you can run them against any provider, including us.
Why coverage maps lie
A coverage map costs nothing to publish. A country can appear on one because a partnership exists on paper, because a single corridor once ran, or because a sub-provider two layers down claims it. None of that tells you what happens to Tuesday's payroll. The questions below are the ones whose answers are hard to fake, because each demands either a number, a named fact, or a written commitment.
The six questions
| Ask | A good answer | A warning sign |
|---|---|---|
| How is coverage counted? | Countries where production transfers actually land | A map with no method or volume behind it |
| Are delivery times measured or promised? | Medians over a stated window, published per rail | “Instant” with no number attached |
| What does funding require? | Fund at execution, from one treasury asset | A prefunded float in every market before you start |
| Do collections and payouts share a balance? | Money collected funds money paid, one ledger, same day | Separate providers and reconciliations per direction |
| How deep do local methods go? | Mobile money by phone number, instant bank rails by name | “Bank transfer” as the only method in wallet-first markets |
| Who regulates the counterparty? | Named licences and safeguarding, in writing | Trust language with no regulator named |
The question that filters fastest
Measured versus promised. A provider that runs real volume can tell you its median delivery time per rail over a stated window, because the number falls out of production data. A provider that cannot is telling you something too. We publish 90-day medians on every rail page and a quarterly report on African payment rails: Nigeria bank payouts over NIP at a 57-second median, every measured mobile money rail under 95 seconds. Ask every provider on your shortlist for the same numbers, over the same window.
Shortlist two providers. Week one: integrate both sandboxes; the time this takes is itself a signal. Week two: send small production volume through both, to your three hardest markets, not your three easiest. Measure delivery medians yourself, force a failure and watch the error handling, and pull the reconciliation exports into your ledger. Decide on your data, not the deck.
How we answer the six
Coverage counted from successful transfers: 18 payout countries, each with its own spec sheet. Delivery medians measured over rolling 90 days and published per rail. Funding at execution from USDC or USDT, no float per market. Collections in 9 African markets settle into the same balance payouts run from. Mobile money addressed by phone number across MTN MoMo, M-Pesa and Airtel Money, plus named instant rails like NIP. Licensed by the Bank of Uganda, registered with FinCEN, authorised for remittances in Kenya, with customer funds safeguarded and segregated. Run every other provider through the same table.
Every rail we run, measured.
18 countries of spec sheets: delivery medians, exact fees and working code per rail.
See the spec sheets