ABA Editorial · Mar 4, 2026 · 12 min read
Safaricom M-Pesa operates approximately 298,890 agents serving 35.82 million customers and KSh 38.29 trillion in FY25 transaction volume. OPay and PalmPay have built comparable agent networks in Nigeria. Uganda's Agent Banking Company supports over 20,000 shared agent outlets. Agents are the bridge between digital money and cash economies. Inside the unit economics, fraud problems, and the reason mobile money cannot work without them.
Digital financial services in Africa run on smartphones, APIs, and cloud infrastructure. But the money itself, the cash that customers deposit at the start of the journey and withdraw at the end, runs through physical agents: small shop owners, kiosk operators, market traders, and dedicated mobile money agents who exchange physical cash for digital balance and back. Without agents, mobile money is just a closed-loop balance that nobody can spend. With agents, mobile money becomes the real payment infrastructure of Sub-Saharan Africa. This is a market report on the agent layer of African fintech, the unit economics that make it work, the fraud problems that threaten it, and why the operators building agent networks are doing some of the most important infrastructure work on the continent.
Safaricom M-Pesa is the pioneer and still the benchmark. According to Safaricom's FY25 (fiscal year ending March 2025) disclosures, M-Pesa operates approximately 298,890 agents across Kenya, serving approximately 35.82 million customers who together generated approximately KSh 38.29 trillion (approximately USD 296.8 billion at fiscal year exchange rates) in transaction volume. That is over 37 billion transactions in a single year processed across a network that is mostly small kiosk operators and neighborhood shopkeepers. The agent economics, the shop rebate per transaction, the float management requirements, the KYC and compliance obligations, have been refined over nearly two decades of operation and remain the template other mobile money operators study.
The critical insight from the M-Pesa story is that agents are not just a distribution channel. They are also the trust layer. A customer who deposits cash with a neighborhood agent does so because they trust the agent personally, because they see the agent every day, and because the agent represents a physical accountability that a pure digital service cannot provide. Early M-Pesa adoption research found that agent trust was the single most important factor in first-time customer conversion. Without reliable agents, mobile money launches in new markets fail regardless of how good the underlying technology is.
The Nigerian mobile money story is different because there is no M-Pesa-equivalent single dominant operator. Instead, Nigeria has multiple operators competing aggressively for agent network scale. OPay (backed by Chinese investors including Opera) has built one of the largest Nigerian agent networks, leveraging aggressive incentives and a super-app product strategy. PalmPay (also Chinese-backed) reportedly passed 35 million users according to coverage of the Nigerian open banking framework, and has built its own extensive agent network. Paga, one of the original Nigerian mobile money operators, continues to operate a significant agent network and has added lending and other services on top.
The Nigerian agent network competition has been intense, occasionally destructive, and always dynamic. Agents frequently operate multiple operators' platforms simultaneously and choose which to promote to customers based on rebate structures. This multi-homing breaks some of the economic assumptions mobile money operators use to justify initial agent recruitment investments, but it also makes the Nigerian agent economy unusually resilient: no single operator's collapse would remove agent access for Nigerian customers.
Covered in an earlier batch of this Market Reports series, the Nigerian Payment Service Bank (PSB) framework explicitly mandates that PSBs must have 25 percent of their agent touchpoints in rural areas. This is a regulatory tool for using bank licensing to force the creation of agent networks in underserved regions. MTN MoMo Nigeria and Airtel SmartCash have both had to invest in agent network expansion partly because of PSB obligations. OPay, which operates under a microfinance bank license rather than a PSB license, does not have the same rural coverage obligation but has built rural agent presence for competitive reasons.
Uganda has taken a distinctive approach to agent network economics. In 2018, Ugandan banks formed the Agent Banking Company, a shared agent infrastructure that now operates over 20,000 outlets nationwide. Rather than each bank building its own agent network from scratch, the banks pool resources into a shared platform and split access according to a negotiated framework. This is the same underlying logic that produces a national payment switch, applied to physical agent infrastructure. The Ugandan approach has been particularly effective at extending agent presence into markets that would not support a single bank's agent investment alone.
In Ghana, GhIPSS operates the Mobile Money Interoperability (MMI) service at the payment layer, and the interoperable agent layer has evolved alongside it. MTN, AirtelTigo, and Vodafone (now Telecel) agents all operate in the same commercial environment, and the competitive pressure plus the interoperability framework have kept agent economics relatively healthy.
Agent networks generate operational risk at a scale that is hard to appreciate from outside the sector. Every agent location is a potential fraud point. Agents can conduct fake transactions to collect rebates. Agents can accept stolen identity documents to onboard customers who will then use the account for fraud. Agents can collude with customers to defeat KYC requirements. The fraud control systems that mobile money operators run against agent networks are among the most sophisticated anti-fraud systems in African financial services, and they have to be, because agent network integrity is the foundation of customer trust.
Safaricom M-Pesa has publicly described machine-learning-based fraud detection systems that actively monitor agent transaction patterns for anomalies. Other operators have developed similar tools. The regtech layer covered in a sibling article in this batch (Smile ID, Youverify, Dojah, Identitypass) is heavily used by mobile money operators to verify customer identities at agent-mediated onboarding, specifically because the agent-mediated channel is where fraud risk concentrates.
The basic mobile money agent unit economics work like this. An agent maintains a float balance, typically funded from their own working capital or from an agent-specific credit line provided by the mobile money operator. When a customer wants to deposit cash, the agent accepts the cash and transfers an equivalent digital balance to the customer's wallet, earning a commission. When a customer wants to withdraw cash, the agent accepts the digital transfer back and hands over physical cash, again earning a commission. The agent's float rotates between physical cash and digital balance as customer demand shifts through the day.
Commission rates are typically a small percentage of transaction value, often tiered so that larger transactions earn proportionally smaller rebates. An agent's profitability depends on volume, float turnover speed, and the mix of deposit versus withdrawal transactions. Good agents run their mobile money business alongside a core retail operation (a shop, a restaurant, a phone accessories stall) and benefit from the foot traffic that mobile money generates. Agents who try to operate mobile money as a standalone business often struggle with unit economics.
Three things. First, how the continued growth of super-apps (OPay, PalmPay, MNT-Halan in Egypt) changes agent network dynamics as customers move more activity to pure digital channels that bypass agents entirely. If customers increasingly deposit and withdraw through embedded channels rather than agents, agent economics get compressed. Second, whether interoperability frameworks like Ghana's GhIPSS MMI and Kenya's post-2022 mobile money interoperability rules change who captures the value in multi-operator agent operations. Third, whether fraud losses from agent networks remain manageable as transaction volumes continue to grow, or whether a major fraud event triggers regulatory tightening that compresses agent commissions below viability.
The longer-term observation is that African mobile money has proved something unique: cash-digital bridging can work at continental scale if the agent economics are right and the regulatory framework is supportive. This is infrastructure that does not exist at the same scale anywhere else in the world, and it is the foundation on which every African consumer fintech covered in this Market Reports series depends. When people write the story of African financial inclusion in the 2020s, the invisible hero is not any specific operator. It is the 298,890 M-Pesa agents, the Nigerian agent networks, the Ugandan shared outlets, and the hundreds of thousands of neighborhood shopkeepers across the continent who exchange cash for digital balance and back, several billion times a year, making the whole system work.
This report draws on Safaricom FY25 investor disclosures and annual report materials (fiscal year ending March 2025); GSMA State of the Industry Report on Mobile Money 2026; TechCabal analysis of Pesalink and Kenyan payment infrastructure (October 2025); Tech In Africa and GhIPSS analyses of Ghanaian interoperability; public materials from OPay, PalmPay, and Paga; and prior batches of this Market Reports series on Nigerian PSB regulatory framework and the Ugandan Agent Banking Company shared infrastructure.