Login Almost every service business that decides to sell into Africa makes the same move first: it turns on its existing card processor, points it at a new set of countries, and waits. Then it looks at the funnel three months later and finds thousands of sessions, hundreds of pricing-page views, and a conversion rate somewhere near zero. The instinct at that point is to blame the market — maybe there just isn’t demand — and quietly deprioritise the region.
That conclusion is almost always wrong, and it is expensive. The demand is there. What is missing is the ability to pay you.
The demand is already there — the payment method isn’t
Start with scale, because the scale is what makes the conversion problem worth solving rather than tolerating. Sub-Saharan Africa processed $1.4 trillion in mobile money transactions in 2025 — roughly two-thirds of the global total of $2.1 trillion. The industry took twenty years to pass its first trillion dollars in annual transaction value, and then four years to double it.
Mobile money value moved through sub-Saharan Africa in 2025 — 66% of the global total. Almost none of it touches a card network.1
That last sentence is the whole problem in one line. This is an enormous, functioning, growing payments economy that your Stripe or Adyen integration cannot see. The money is moving. It is moving through mobile wallets, agent networks and USSD strings, and a checkout that only offers Visa and Mastercard is standing in a different building entirely.
The failure is quiet, which is what makes it dangerous. A card decline at least produces an error you can go and read. A buyer who opens your pricing page, scans the payment options, sees nothing they can use, and closes the tab produces nothing at all — no decline code, no failed payment, no support ticket. It shows up in your dashboard as ordinary bounce, indistinguishable from disinterest. You will read it as a marketing problem and spend the next quarter rewriting your landing page.
Pull your last 90 days of traffic and filter by African countries. If you’re seeing meaningful sessions with a near-zero conversion rate — and particularly if you see people reaching checkout and abandoning at the payment-method step — you don’t have a demand problem. You have an acceptance problem, and it is fixable.
Why the card checkout fails, in two layers
Merchants tend to assume the fix is a better card processor, a smarter retry cascade, or a local acquirer. Occasionally that helps at the margin. Mostly it doesn’t, because the failure happens in two places and only one of them is inside your checkout.
Layer one: the card doesn’t exist
The majority of your prospective African buyers do not hold a card that works for cross-border online purchases. Some hold no card at all. Many hold a domestic debit card that is not enabled for international e-commerce, or is enabled but capped at a limit that makes your annual plan impossible. This is not a fringe segment you can round off — in most of these markets it is the mainstream.
What that means practically: the checkout is not converting badly, it is not being attempted. There is no funnel to optimise. The customer arrived willing to pay and left because you did not offer a method they could use.
Layer two: when a card does exist, it often gets declined anyway
For the minority who do hold a usable card, cross-border authorisation is genuinely hostile. Issuers in these markets treat foreign online merchants as elevated risk and decline aggressively. 3-D Secure challenges time out on unstable mobile connections. The customer typically gets no useful explanation, tries once more, and gives up. From their side it looks like your product doesn’t want their money.
Optimised for a customer who isn’t there
- Most buyers hold no cross-border-capable card
- Cross-border authorisation rates are weak and issuer-dependent
- 3-D Secure fails on poor connections
- Failures are invisible — they look like bounce, not decline
- No retry logic recovers a payment that was never attempted
Optimised for how the money actually moves
- Reaches buyers who have a wallet balance and no card
- Customer authorises with a PIN prompt on their own phone
- No card network, no cross-border issuer decline
- Push-confirmed — the payment either completes or clearly fails
- Works on the device and the rails people already use daily
Why do card payments fail so often in Africa?
How mobile money actually works — and why it changes your product
If you have only ever built against card rails, mobile money will feel structurally backwards, and that instinct is worth taking seriously. The difference is not cosmetic. It changes what you can and cannot do to a customer’s money, and it changes your billing design.
Cards are pull-based. The customer hands you a credential once. From then on, you initiate. You can charge the card at 3am on a Tuesday without the customer touching anything. Your entire subscription business is built on that assumption, whether or not you’ve ever articulated it.
Wallets are push-based. There is no stored credential you can silently pull against. The customer confirms each payment with a PIN, on their own phone. You request; they approve. That’s the whole model, and it does not bend.
Customer picks their wallet at checkout
They select M-Pesa, MTN MoMo, Orange Money, Airtel Money, Wave — whichever is dominant in their market — and enter the phone number attached to it. No card number, no CVV, no billing address.
A prompt lands on their handset
An STK push or equivalent appears on the phone showing the merchant name and the amount. On feature phones this is a USSD string. Either way, it arrives on the device they’re already holding.
They authorise with a PIN
One PIN entry. This is the security step, and it replaces 3-D Secure entirely. There is no redirect to an issuer page that may or may not load on a 3G connection.
Funds move and you get a webhook
The value leaves their wallet balance and lands with the acquirer. Your system gets a definitive confirmation and provisions access. Typically the whole loop takes seconds.
Revenue settles out to you
The collected local currency is converted and moved to you on your agreed schedule, in the currency you asked for. This is the step most merchants underestimate — see section 06.
The consequence is that mobile money is dramatically more reliable at the moment of payment — there is no issuer sitting in the middle looking for a reason to say no — and dramatically less flexible afterwards. You gain acceptance. You give up the ability to bill someone silently. If you are running a subscription, that trade has real product implications, and we come back to it in section 08.
Chargebacks work differently too. Card chargebacks are a customer-initiated, scheme-governed process that can hit you months after a payment. Mobile money disputes are typically resolved through the operator and the acquirer, on shorter timelines, with less automatic reversal in the customer’s favour.
This is generally good news for a service merchant, but it is not an absence of risk — it means your recourse, and your customer’s recourse, run through a different set of rules. Do not assume the card playbook transfers.
Market by market: what you can actually accept
“Africa” is not a payment market, and treating it as one is the second-most-common mistake after the card-only checkout. Fifty-four countries, different regulators, different dominant wallets, different FX regimes. A stack that works beautifully in Nairobi may collect nothing at all in Dakar.
Here is the honest shape of it for a service seller — not an exhaustive map, but the markets where demand and rails are both mature enough to be worth a launch.
MTN MoMo dominates. Interoperability between wallets is unusually good, which simplifies things. Cards exist among urban professionals but won’t carry your volume.
| Market | Dominant rails | Card viability | What to know |
|---|---|---|---|
| 🇰🇪 Kenya | M-Pesa, Airtel Money | Low | The market that invented mobile money still runs on it. M-Pesa holds the large majority of wallet share. If you sell into Kenya without M-Pesa, you are not selling into Kenya. |
| 🇬🇭 Ghana | MTN MoMo, Telecel Cash, AirtelTigo | Mixed | MTN MoMo dominates. Interoperability between wallets is unusually good, which simplifies things. Cards exist among urban professionals but won’t carry your volume. |
| 🇳🇬 Nigeria | Bank transfer, USSD, cards, wallets | Mixed | The outlier. Nigeria is bank-transfer-first, not wallet-first. Instant transfer is culturally default. FX access is the real constraint here, not acceptance. |
| 🇸🇳 Senegal | Wave, Orange Money, Free Money | Low | Wave has taken serious share on aggressive pricing. XOF is pegged to the euro, which makes FX unusually predictable for a merchant. |
| 🇨🇮 Côte d’Ivoire | Orange Money, MTN MoMo, Wave | Low | Fragmented across three serious players — you need all of them, not one. Also XOF, so the same euro peg applies. |
| 🇨🇲 Cameroon | MTN MoMo, Orange Money | Low | A clean duopoly. Cover both and you have effectively covered the market. XAF, also euro-pegged. |
| 🇹🇿 Tanzania | M-Pesa, Tigo Pesa, Airtel Money | Low | Genuinely interoperable wallets, high penetration. Similar behaviour to Kenya but with more distributed share. |
| 🇺🇬 Uganda | MTN MoMo, Airtel Money | Low | Mobile money tax has been a live policy issue and affects behaviour at low ticket sizes. Price accordingly. |
| 🇿🇦 South Africa | Cards, EFT, instant EFT | High | The exception that proves the rule. Card and bank penetration are high; standard card processing genuinely works. Do not generalise from South Africa to the continent. |
Two things fall out of that table. First, South Africa is not representative, and a founder whose only African experience is Johannesburg will draw exactly the wrong conclusion about Lagos or Abidjan. Second, Nigeria is its own animal — the acceptance question is largely solved by instant bank transfer, and the hard part is getting your revenue back out through an FX regime that has been volatile for years.
Which African markets should a service business start with?
Building the stack: three routes, three different bills
Once you accept that you need wallet acceptance, you have to decide how to get it. There are three real options, and the marketing material for each will not tell you what it actually costs you.
Route one: register locally and integrate direct
Set up an entity in each country, open a local bank account, get your own merchant code with each mobile network operator, and build against their APIs. This is the route people default to when they think of “doing it properly.”
It gives you the best possible unit economics per transaction and full control of the relationship. It also gives you company registration in every market, local directors or resident agents where required, tax and regulatory reporting obligations, one integration per operator per country, and a launch timeline measured in quarters rather than weeks. For a business selling a service — as opposed to one building a payments company — the arithmetic almost never works until a single market is producing serious, sustained volume.
Route two: stack aggregators
Use a global PSP, which connects to a regional aggregator, which connects to a local aggregator, which finally touches the operator. It’s fast to launch and requires almost no thinking, which is exactly why it’s the default.
The problem is that every party in that chain takes a cut, and the cuts are not additive in a way you can easily see. You are quoted a headline rate by the party at the top. Underneath it sit a settlement FX spread you were not shown, a fixed fee per transaction, and a margin at each hop. Merchants routinely discover their real all-in cost is several points above the number in the contract, and by the time they’ve reconciled it they’ve already built their pricing around the wrong figure.
Total margin loss observed across stacked aggregator chains on some African corridors — versus a quoted headline rate that looked far lower.2 The number in your contract is not your cost.
Route three: one intermediary holding the local relationships
A single partner sits between you and the operators. They hold the acquiring relationships, the licences and the local presence. You integrate once. They handle the wallet-by-wallet mechanics, the settlement, and the FX.
This is the route that works for most service businesses, for an unglamorous reason: it collapses eleven integrations and eleven regulatory conversations into one, without inserting three margin-taking parties into the chain. The thing to interrogate — and this is the whole diligence process in one question — is how many parties actually sit between the partner and the operator. Some “direct” providers are aggregators wearing a nicer landing page. Ask, and ask specifically.
“Do you hold the acquiring relationship with the operator yourself, or through a partner?” Then: “What is the all-in cost after FX, on a $50 payment in Kenya, settled to USD?” A provider who can answer both without a follow-up call is worth talking to. One who can’t is a layer in a chain, and you will pay for it every month.
Getting the money out — the part nobody plans for
Acceptance is the half of the problem everyone focuses on. Settlement is the half that quietly determines whether the business works.
You have now collected Kenyan shillings, Ghanaian cedis, and West African CFA francs. None of that is money you can spend. It has to be converted and moved to wherever your business actually banks, and every step of that has a cost and a delay.
The variables that decide whether this is painless or a monthly ordeal:
- FX rate and spread. The single largest hidden cost in African payments. A partner quoting you 2.5% acceptance with a 4% FX spread is charging you 6.5%, and only one of those numbers appeared in the pitch deck. Always ask what rate you’re getting and what the reference rate is on the day.
- FX rate and spread. The single largest hidden cost in African payments. A partner quoting you 2.5% acceptance with a 4% FX spread is charging you 6.5%, and only one of those numbers appeared in the pitch deck. Always ask what rate you’re getting and what the reference rate is on the day.
- Settlement frequency. Daily, weekly, or on a threshold. Faster settlement means less of your working capital parked in a currency you don’t control. If a provider is vague about frequency, that vagueness is the answer.
- Rolling reserve. Some providers hold a percentage of your revenue for a rolling period against dispute risk. Perfectly legitimate for a genuinely high-risk merchant — but if you’re a straightforward SaaS or edtech business, a 10% rolling reserve for 180 days is a working-capital tax you should negotiate hard on or walk away from.
- Regulatory friction on the way out. This is the Nigeria problem specifically, and it is not a partner-selection issue you can solve by shopping around — it is a feature of the local FX regime. Know it going in.
Can I get my revenue out of local currency?
What it actually costs
Realistic all-in pricing for mobile money acceptance sits in the region of 2.9% to 3.9% per successful payment, varying by market, volume, and settlement currency. That is higher than you’d pay for a domestic card transaction in a mature market, and merchants sometimes bounce off the number on principle.
That reaction is a category error. You are not comparing 3.4% against 1.9%. You are comparing 3.4% against zero revenue, because the alternative is a checkout that does not collect from these customers at all. The relevant question is not “is this cheap” — it is “does the revenue this unlocks exceed the cost of collecting it,” and for a service business with any margin at all, it does, comfortably.
What you should be aggressive about is not the headline rate. It’s the things underneath it:
| Line item | What to accept | What should worry you |
|---|---|---|
| Acceptance rate | ~2.9–3.9%, quoted per market | A single blended rate across every country — it hides where you’re being overcharged |
| FX spread | Quoted explicitly, against a named reference rate | “Market rate” with no reference, or no answer at all |
| Fixed fee | Small or none; matters most at low ticket | A fixed fee that quietly makes sub-$5 transactions uneconomic |
| Rolling reserve | 0% for a clean, low-dispute service merchant | 10%+ over 180 days with no path to reducing it |
| Settlement delay | Daily to weekly, stated in the contract | “Typically” — an unwritten timeline is not a timeline |
| Setup / monthly | Zero, or clearly justified | Meaningful minimums before you’ve validated the market |
What does mobile money acceptance cost?
Recurring revenue without a stored card
If you sell a subscription, this section is the one that will cost you money if you skip it.
Your renewal logic assumes a stored credential you can pull against. Wallets don’t give you one. Ship your existing billing engine into a wallet market unchanged and you will get renewal failure that looks exactly like churn — and you will misdiagnose it, because every dashboard you own will report it as churn. You’ll go and interview the customers, and they will tell you they liked the product. They just never got billed.
There are two patterns that work, and they are genuinely different products.
Pre-funded balance
The customer tops up a balance held with you. Renewals draw down against it automatically, no prompt required. When the balance runs low, you notify them and they top up again.
- Renewals become silent and reliable again
- You hold customer funds — treat this as a real obligation, not a float
- Top-up prompts must be early and unmissable
- Best fit for frequent, low-ticket usage
Scheduled prompt
On the renewal date, a payment request lands on the customer’s phone. They confirm with their PIN. Simple, transparent, and it never touches money you’re holding on their behalf.
- No custody of customer funds
- Every renewal is an active re-consent — churn is honest
- Timing matters enormously — a prompt at 2am is a lost renewal
- Best fit for higher-ticket, lower-frequency plans
Whichever you pick, the operational discipline is the same: a renewal prompt that arrives at a bad moment is not a failed payment, it is a churn event you caused. Send it when the customer is awake and likely to have a balance. Retry with judgement rather than on a fixed cron. Give them a way to top up before the renewal, not after it fails.
How do subscriptions work without cards?
Six expensive mistakes
These are the ones we see repeatedly, in roughly the order they cost the most.
Treating Africa as one market
Launching a single wallet integration and calling the continent covered. The wallet that owns Kenya is irrelevant in Senegal. The rails that carry Nigeria are bank transfers, not wallets at all.
Optimising the card funnel
Spending a quarter on retry cascades, smart routing, and 3DS tuning for a customer base that overwhelmingly does not hold a usable card. You are polishing a door nobody is walking through.
Comparing headline rates
Choosing a provider on the acceptance percentage in the pitch deck, then discovering the FX spread on settlement is doing more damage than the acceptance fee ever was.
Shipping billing logic unchanged
Running card-shaped renewal logic against push-based rails, then reading the resulting failures as churn and rebuilding the product to fix a problem that was never in the product.
Pricing in dollars only
Showing a USD price to a buyer who thinks in shillings, cedis or CFA. Every conversion the customer has to do in their head is friction, and unclear FX at checkout reads as a trap.
Ignoring the phone the customer is holding
A checkout that requires a stable connection, a modern browser and a large screen, in markets where the median session is mobile, intermittent and data-conscious.
The launch checklist
If you’re taking a service to market in Africa this quarter, this is the shortest honest version of what has to be true before you turn it on.
Before you launch
- ✓ You’ve picked specific markets, not a region. Named countries, chosen from where your traffic already goes to die.
- ✓ You cover the dominant rails in each of them. Not “a wallet” — the wallets, plus bank transfer where that’s the default.
- ✓ You know your all-in cost, after FX. One number, per market, in writing.
- ✓ You know how many parties sit between you and the operator. You asked directly and got a direct answer.
- ✓ Settlement currency, schedule and reserve are in the contract. Not in an email, not “typically.”
- ✓ Your renewal mechanics survive push-based payments. Pre-funded balance or scheduled prompt, designed deliberately.
- ✓ Checkout shows local currency at a rate the customer can trust.
- ✓ You’ve tested on a real mid-range Android on a bad connection. The whole flow, end to end.
- ✓ Support can answer “my payment failed” in the customer’s context. Wallet errors are not card errors, and your macros know the difference.
The market isn’t the problem
The businesses that struggle to sell services into Africa are, with striking consistency, the ones that brought a card-shaped payment stack to a wallet-shaped market and concluded from the results that the demand wasn’t real. The demand is real. A $1.4 trillion mobile money economy is not a hypothesis.
What it requires is the boring, specific work: pick the markets, cover the actual rails, know your all-in cost after FX, redesign your renewals for push-based payments, and test on the phone your customer is genuinely holding. None of that is glamorous, and all of it is the difference between a region that converts and a region you eventually stop reporting on.