M-Pesa: The Pilot That Almost Wasn't a Payments Product

Published by Prakashak Saturday, October 3, 2026

Part of our series on financial inclusion around the world: the story behind each initiative, what works, and what doesn't.

The scene

Imagine a worker in Nairobi who wants to send part of the week’s earnings to family in a village several hours away. In the past, the money might have travelled with a bus driver, a friend, or a relative. Now the worker can use a basic mobile phone to send electronic value. At the other end, a local shop that acts as an M-Pesa agent can convert that value into cash.

The transaction is simple on the surface. Behind it sits a network of mobile connectivity, customer accounts, agents with enough cash and electronic float, transaction records, and rules for resolving mistakes. M-Pesa’s story is not just about putting money on a phone. It is about building that entire system around a problem people already needed to solve.

The story behind it

M-Pesa began as a very different product from the one that later became famous. In the early 2000s, researchers working on financial inclusion had observed people transferring mobile airtime as a rough substitute for money. The idea suggested that mobile networks might support transfers of monetary value as well as voice and text services.

Vodafone received support through the UK Department for International Development’s Financial Deepening Challenge Fund. The project was developed with Safaricom, Vodafone’s Kenyan mobile-network affiliate, and a microfinance partner. The pilot began in October 2005 and tested whether microfinance customers could use a phone-based system to make loan repayments. The initial concept was useful for reducing the need to travel to make regular repayments, but it was not the use that ultimately drove mass adoption. Pilot learning and customer behaviour pointed towards a broader need: sending money to other people.

Safaricom launched M-Pesa commercially in Kenya in March 2007 with the message “Send Money Home”. The service was designed around domestic transfers, while also supporting cash deposits, withdrawals and other transactions. In effect, the product found a larger purpose through use: customers were not simply adopting a new way to repay loans; they were adopting a practical way to move money between people.

Ownership of the brand later changed. In April 2020, Safaricom and Vodacom completed a joint-venture acquisition of the M-Pesa brand, product development and support services from Vodafone Group. Vodafone had helped develop the original service, but it was no longer the owner of the M-Pesa brand after that transaction.

Regulation was another important part of the story. M-Pesa was not launched as a conventional bank account. The Central Bank of Kenya considered how the service should operate under the country’s existing legal and supervisory framework. The case illustrates how regulators can face a new model that does not fit neatly into familiar categories. It also shows why innovation needs a clear framework for safeguarding customer funds, overseeing providers and handling complaints.

How it works, in plain language

In its basic form, M-Pesa links a customer’s mobile number to an electronic-money account. A customer does not need a traditional bank branch for every transaction. Instead, authorised agents provide the physical bridge between cash and electronic value.

  1. Deposit: A customer visits an M-Pesa agent and gives the agent cash. The agent processes a deposit to the customer’s M-Pesa account. The customer receives a confirmation message.
  2. Transfer: The customer uses the phone menu or application to enter the recipient’s number and the amount. The system displays transaction details for confirmation. Once the transfer is completed, both parties receive a notification.
  3. Withdrawal: The recipient visits an agent, provides the required transaction details and follows the service’s authentication process. The agent pays out cash and the electronic value is deducted from the recipient’s account.

The exact screens, limits, identification requirements and fees can change over time and may vary by transaction type. Customers should check the current tariff and follow the instructions shown by the service rather than rely on old examples.

Why the agent is the heart of the model

An electronic balance is useful only if people can turn it into cash when they need to—or use it to pay someone who accepts it. M-Pesa’s agents made that possible in neighbourhoods where a bank branch might be distant or inconvenient. Many agents were existing small businesses such as airtime sellers and local shops. They added cash-in and cash-out services to their everyday operations.

Agents need two forms of liquidity. They need cash to pay customers who withdraw, and electronic float to process deposits and other transactions. A shop with plenty of cash but too little e-float may be unable to accept a deposit; an agent with plenty of e-float but insufficient cash may be unable to complete a withdrawal. Agents therefore manage a constantly changing balance between the two.

Agents earn commissions under the provider’s commission arrangements. The commission gives a shop a reason to offer the service, but the business still has costs: working capital, time, connectivity, security, record-keeping and the risk of being unable to serve a customer at a busy moment. The network succeeds when agent economics work well enough to keep outlets available and when customers can find an agent with the right balance at the time they need one.

What worked

1. A service built around an existing need

M-Pesa addressed a familiar problem: moving money between people who were not necessarily near one another and who might not share access to a bank. “Send Money Home” communicated a concrete use rather than asking customers to understand an abstract financial technology. The early shift from loan repayment to general person-to-person transfers is a reminder that customer behaviour can reveal a product’s real value proposition.

2. A distribution network people could reach

Safaricom already had a substantial mobile customer base and a network of airtime retailers. The service could build on those relationships rather than create an entirely new physical distribution system. Agents brought the service into ordinary commercial spaces and made cash conversion possible beyond bank branches.

The network expanded rapidly. Safaricom’s FY2026 reporting for the year ended March 2026 put Kenya’s one-month-active M-Pesa customer base at about 41 million and its agent network at more than 333,000. These are company-reported activity and network measures, not counts of unique people who were previously unbanked. They show the scale reached by the platform, but scale alone does not establish the quality or depth of financial inclusion.

3. Evidence of changes in household welfare

A widely cited study by Tavneet Suri and William Jack, published in Science in 2016, estimated that increased access to M-Pesa raised per-capita consumption and lifted approximately 194,000 Kenyan households—about 2% of Kenyan households—out of poverty. The study reported stronger effects among female-headed households. It linked the observed changes to greater financial resilience and saving, as well as changes in work, including some women moving from agriculture into business and retail activities.

The unit matters: the headline estimate refers to households, not 194,000 individual people. The authors estimated the effect of increased access to the agent network, using changes in proximity to agents, rather than simply comparing people who chose to use M-Pesa with those who did not. The result is important evidence, but it should not be turned into a claim that every user benefited equally or that mobile money alone explains Kenya’s poverty trends.

4. Convenience that complemented, rather than simply replaced, banking

Early research found that many initial M-Pesa users already had bank accounts. For them, the service complemented existing financial services. It could make transfers faster or more convenient without replacing a bank account, savings product or other formal service. This is an important distinction: financial inclusion is not only about reaching people who have no account; it can also involve making everyday financial activity more accessible and useful.

What didn't, or is still unresolved

1. Dependence on a dominant provider

M-Pesa’s close association with one mobile operator helped create a consistent service and a large network. It also concentrated important functions—customer access, transaction infrastructure, agent relationships and product rules—within a powerful provider ecosystem. Such concentration raises questions about competition, pricing, interoperability, resilience and the options available to customers if a provider experiences disruption or changes its terms.

These are structural questions, not proof that every customer is overcharged or that the service is unreliable. They are reasons to examine the market’s rules and performance alongside its reach.

2. Access does not guarantee positive outcomes in every use

Mobile money makes it easier to move money for many purposes, including purposes that may create financial or social risks. Research using Kenya’s 2021 FinAccess Household Survey found an association between active mobile-money use and a higher likelihood of gambling, with the relationship varying across groups and with frequency of use. This does not mean that mobile money necessarily causes harmful gambling in every case. It does show that payment access is not, by itself, a measure of welfare. The same convenient rails can support useful transfers and transactions with potential harms.

3. A successful model is not automatically portable

M-Pesa grew in a particular environment: high mobile-network reach, an established telecom operator, a large need for domestic transfers, accessible retail outlets and a regulatory approach that permitted a non-bank-led model to develop. Other countries may have different banking systems, identity infrastructure, agent economics, consumer habits, market structures and rules.

India’s experience is a useful comparison. Vodafone and ICICI Bank announced an M-Pesa service in 2012 and launched it in 2013. The Indian offering combined a Vodafone mobile wallet with an ICICI Bank mobile-money account and used Vodafone’s distribution reach. It was initially offered in parts of eastern India, including Kolkata, Bihar and Jharkhand. Vodafone M-Pesa’s Indian wallet operations were discontinued in 2019 as its RBI authorisation was due to expire. That outcome is not a controlled test proving that the Kenyan model could never work in India. It does, however, show that transferring a brand and broad concept does not reproduce the market conditions, network effects or customer behaviour that supported the Kenyan service.

The view from the ground

For customers, the value of M-Pesa is often measured in saved time, reliable access and the ability to send or receive money without arranging a physical hand-off. A 2009 survey by FSD Kenya, CGAP and the Central Bank of Kenya found that many surveyed users considered M-Pesa faster, more convenient, safer and cheaper than the money-transfer alternatives they had used before. The survey also examined delays, losses, agent experiences and customer care. It is useful early evidence, but it should not be mistaken for a description of every customer’s experience today.

The agent’s experience is more operational. An agent may need to rebalance cash and e-float, especially around paydays, market days or other periods of high demand. A queue can form when one side of the agent’s liquidity runs low. Commissions may generate additional income for a shop, but transaction volume is not the same as profit: the agent must account for working capital, operating costs and the time spent serving customers.

Errors and disputes are also part of a real payment system. A customer may enter the wrong recipient number, experience a delayed confirmation, or need help after a transaction that appears to have failed. A mistaken transfer or suspected fraud can be especially stressful when the amount is significant to a household. Clear receipts, accessible customer support, timely investigation, understandable reversal rules and effective escalation to the relevant provider or regulator are therefore core parts of inclusion—not optional extras.

Customers should verify recipient details before confirming a transfer, keep transaction messages, protect their PIN and use official support channels when something goes wrong. They should not share a PIN or one-time authentication code with someone claiming to be an agent or customer-care representative.

The India lens: different starting assets, different architecture

Kenya’s M-Pesa and India’s financial-inclusion infrastructure share a broad objective: making payments and other financial services more accessible. Their institutional starting points, however, are different.

M-Pesa is telecom-led. In Kenya, Safaricom’s mobile network, customer relationships and retail distribution formed the foundation of the service. The mobile number and M-Pesa account became a familiar way to address transfers, while agents connected electronic value with cash. A telecom operator played a central role in the customer-facing network and service delivery, within the regulatory framework applied in Kenya.

India’s JAM and UPI ecosystem is more distributed across institutions and public infrastructure. Jan Dhan accounts provide bank-account access; Aadhaar can support identity verification and certain authentication or benefit-delivery processes; and UPI enables interoperable bank-account payments through participating banks and payment applications. These components are related but not interchangeable. Aadhaar is an identity system, not a payment network or bank account. UPI is a payment system, not an identity database. A Jan Dhan account is a bank account, not a mobile-money wallet.

This difference affects who does what. In the M-Pesa model, the telecom operator and its agent network are central to the service experience. In India, banks hold customer accounts, UPI connects participating institutions and applications, and Aadhaar may be used in specific identity or authentication workflows. Business correspondents and other assisted-service channels can help customers who need cash or in-person support. The customer may therefore interact with different organisations for account access, payment initiation, authentication and complaint resolution.

Customer verification also follows different arrangements. M-Pesa’s onboarding has been tied to applicable Kenyan identification and registration requirements. In India, bank account opening is governed by banking KYC requirements; Aadhaar may be one permitted identity or address document or may be used in specific permitted processes, but it is not the only route to every bank account or payment. UPI transactions generally rely on the customer’s bank-linked setup and payment authentication rather than requiring Aadhaar for each ordinary payment.

Complaint handling reflects the architecture. An M-Pesa user typically begins with the service provider’s customer-care process, with escalation routes available under Kenyan rules. In India, the correct first point of contact depends on the issue: the bank, payment application, UPI participant or other relevant service provider may need to investigate. Separate regulatory and grievance mechanisms apply depending on the product and institution. A customer should receive a clear route to raise a complaint and know which entity is responsible for resolving it.

The comparison is not a contest between one technology and another. It is a reminder that infrastructure is shaped by institutional roles. A model that works through a telecom’s customer base and agents may not transfer directly to a country where bank accounts, interoperable payment rails, identity systems and assisted banking channels already play substantial roles. The practical question is not whether India should copy M-Pesa, but which design lessons—agent liquidity, simple user journeys, reliable confirmations, accessible support and strong consumer protection—can improve services in India’s own system.

Four questions we ask of every initiative

  1. Who drove it? Vodafone and Safaricom developed the service, with donor-backed pilot support and a Kenyan regulatory environment that allowed the model to proceed. The provider’s existing mobile distribution network was a major starting asset.
  2. How were people onboarded and identified? Customers registered for the service under the applicable Kenyan requirements and used a mobile-linked account. Agents helped make the service accessible, but onboarding still depended on meeting identification and registration rules. Access through a phone did not mean that identity checks or account safeguards disappeared.
  3. Did people keep using it after signing up? The growth in active customers, transaction activity and agent coverage indicates that M-Pesa became an everyday service for many users. The 2016 Suri and Jack study also examined the consequences of expanded access, not merely registrations. Still, headline customer counts do not reveal how frequently every person uses the service, what share of activity is essential, or whether usage produces equal benefits across groups.
  4. What happened to trust, fraud and complaint handling? Trust depended on successful transactions, agent availability, understandable charges, secure authentication and the ability to resolve errors. Early user-and-agent research documented concerns involving delays, losses, fraud and customer support, and recommended better agent training, complaint monitoring and consumer information. Mobile money can also be used for transactions with harmful consequences, including gambling-related activity. A mature inclusion assessment therefore has to examine safeguards and remedies alongside convenience and reach.

What the M-Pesa story tells us

M-Pesa did not become a major payments service simply because a mobile phone could carry a transaction. Its development combined an existing customer need, a telecom operator’s reach, a dense agent network, a product that adapted to how people actually used it, and a regulatory approach that made room for a new model.

Its experience also sets limits on what success figures can tell us. Large customer and agent networks show reach. Research on household consumption and occupational choices provides evidence about welfare effects. Neither, by itself, answers every question about affordability, service quality, competition, fraud, customer recourse or the distribution of benefits.

For India and other countries, the transferable lesson is less about copying a particular platform and more about asking practical questions: Can people access the service where they live and work? Can agents remain liquid? Are charges clear? Does the system work across providers? Can users recover from mistakes? And is there credible evidence that access improves people’s financial lives?

Those questions help distinguish a widely adopted payment product from a financial-inclusion initiative that is accessible, dependable and beneficial over time.

Sources and further reading

 Disclaimer & Regulatory Disclosure

Educational content only, not financial advice.

Examples are illustrations, not recommendations. aadhaarpay.com and its author(s) are not affiliated with UIDAI, NPCI, or any government body or bank, and are not SEBI-registered Investment Advisers, Research Analysts, or Stock Brokers. We provide no payment, banking, or Aadhaar services. Schemes and figures change; please verify with official sources. Full disclaimer →