Nick Hughes and Susie Lonie

M-PESA: Mobile Money for the “Unbanked”

Turning Cellphones into 24-Hour Tellers in Kenya

In March 2007, Kenya’s largest mobile network operator, Safaricom (part of the
Vodafone Group) launched M-PESA, an innovative payment service for the
unbanked. “Pesa” is the Swahili word for cash; the “M” is for mobile. Within the
first month Safaricom had registered over 20,000 M-PESA customers, well ahead
of the targeted business plan. This rapid take-up is a clear sign that M-PESA fills a
gap in the market. The product concept is very simple: an M-PESA customer can
use his or her mobile phone to move money quickly, securely, and across great dis-
tances, directly to another mobile phone user. The customer does not need to have
a bank account, but registers with Safaricom for an M-PESA account. Customers
turn cash into e-money at Safaricom dealers, and then follow simple instructions
on their phones to make payments through their M-PESA accounts; the system
provides money transfers as banks do in the developed world. The account is very
secure, PIN-protected, and supported with a 24/7 service provided by Safaricom
and Vodafone Group.
The project faced formidable financial, social, cultural, political, technological,
and regulatory hurtles. A public-sector challenge grant helped subsidize the invest-
ment risk. To implement, Vodafone had to marry the incredibly divergent cultures
of global telecommunications companies, banks, and microfinance institutions
–and cope with their massive and often contradictory regulatory requirements.
Finally, the project had to quickly train, support, and accommodate the needs of

Two authors have written this case study, each taking up one part of the story. The first
section is written by Nick Hughes, a Vodafone executive who started this project in
2003 and now heads a mobile payments team with the task of growing the business
globally. Nick led the efforts to get Vodafone focused and supportive at the very senior
levels of the company in what seemed at the outset to be a fringe project.
The second section is written by Susie Lonie, an m-commerce expert who was
brought into Kenya to work through the detailed design phase and project manage the
overall delivery of the service from pilot into commercial operation. Susie describes the
day-to-day obstacles that needed managing, many of which could never have been
foreseen at the outset.

© 2007 Nick Hughes and Susie Lonie

innovations / winter & spring 2007 63
Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 Nick Hughes and Susie Lonie

Figure 1. M-PESA screenshot.

Source: Image courtesy of FDCF; photographer: Tom Lee.

customers who were unbanked, unconnected, often semi-literate, and who faced
routine challenges to their physical and financial security. We had no roadmap, but
created solutions as we went and persevered when a pilot slated to take several
months took almost two years.
Why and how does a telecom company like Vodafone start a banking project
like this? It’s not part of Vodafone’s core business; it was not developed in a core
market (Kenya is a relatively small market in Vodafone’s terms); and it has little to
do with the voice or data products that drive Vodafone’s revenue streams. Telecom
companies are young and fast moving; banks are old, traditional, conservative, and
slow moving. And, having generated the concept and decided at the corporate level
to implement it, what are the practical issues that need managing to get the proj-
ect off the ground and into commercial operation? These are the questions we seek
to answer here.

STAGE 1. MILLENNIUM DEVELOPMENT GOALS, VODAFONE, AND

PUTTING THE DEAL TOGETHER
AUTHOR: NICK HUGHES

The backdrop to our case study lies in the development debate. As part of the
Millennium Development Goals,1 many of the world’s leading nations have com-

64 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

mitted to reduce poverty by 50% by 2015.

Traditionally the realm of state organizations, donor agencies and nongovern-
mental organizations, development is today understood to be unachievable with-
out the engagement of the private sector. There are now more than 2 billion mobile
phone users world-wide and for the vast majority of people the first phone call
they ever make will now be on a
mobile device. When I joined
Vodafone in 2001, I welcomed
the opportunity to work in a
Getting cash into the hands of
sector that is witnessing one of people who can use it is limited
the fastest adoptions of a new
technology the world has ever
on the supply-side rather than
seen. Vodafone is a relatively demand-side; there is no
young, technology-based serv-
ice organization that is keen to shortage of funds, but it’s the
proactively manage its impact ability to move money from the
on society. Part of my initial job
in Vodafone was to help it sender to the receiver that is the
understand its role in address- stumbling block.
ing issues like the Millennium
Development Goals.
In my role at Vodafone, I
had become aware of a promising approach to tackling sustainable development,
one in which I was sure Vodafone could significantly and legitimately play a role.
This approach is simple: access to finance facilitates entrepreneurial activity. In
turn this creates wealth through economic activity, job creation, and trade. There
has been much positive discussion in recent years about donor agencies seeking
new ways to deliver funds to those who need it most, directly and in a more effi-
cient manner, so that the capital is productively deployed. Witness the number of
large, private sector-derived Foundations and Trusts that are addressing poverty
alleviation through enterprise, bringing the rigors of an investment-style approach
to the debate. These range from the Gates and the Soros Foundations to new social
investment funds. At the core of these initiatives is a willingness to find more effec-
tive ways of delivering assistance—a hand up, not a hand out.
Getting cash into the hands of people who can use it is limited on the supply-
side rather than demand-side; there is no shortage of funds, but it’s the ability to
move money from the sender to the receiver that is the stumbling block. Since the
creation of money, the ability to move it from A to B—the so-called “velocity of
money”—has been a fundamental cornerstone of economic activity. But the issue
is exactly how money transfer is made to happen in an emerging market where the
infrastructure is poorly developed and where very few people have or even want
bank accounts. Under such circumstances, moving cash is risky, expensive, and
slow. Enter telecom network operators, who can adapt mobile technology to deliv-
er financial services in a fast, secure and low cost way, especially in developing parts

innovations / winter & spring 2007 65

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

of the world where microfinance institutions have begun to spread and begun to
build an infrastructure.

Getting Funding
I started my M-PESA journey at the World Summit for Sustainable Development
in 2003. After spending an afternoon contributing to a debate about how private
sector organizations are driven by short term goals and thus don’t typically address
long-term sustainable development, I was approached by a representative of the
U.K. government who controlled a challenge fund project set up by the
Department for International Development (DFID). Our discussion centered on
the following: Private sector organizations such as Vodafone are legally bound to
use their shareholders capital to achieve the best returns. But many organizations
use internal competition to allocate funds to their projects, and this competition is
based on potential returns on investment. As a result, any initiatives that relate to
the development agenda usually get squeezed out. Without wishing to overgener-
alize, often the only place within an organization where the development agenda
can ascend is in departments that are more concerned with stakeholder engage-
ment, government relations, policy debate and corporate reputation. How could
firms raise executive-level interest and get funding to develop products that will be
non core and long term but do have some sort of sustainable development theme?
One angle could be to position such projects in the Research & Development
(R&D) department. This would work in many sectors where new products take a
long time to reach market, but many technology-based companies—and Vodafone
is no exception—tend to keep R&D focused on the technology rather than the
marketplace. Financial services in emerging markets are not about new technolo-
gy; in fact, even at the early concept stage we expected to use a very basic applica-
tion of mobile communication, called SMS (or text messaging), certainly not the
sexiest aspect of mobile technology, especially in the core European markets where
java applications, 3G and smart phones were all in vogue. This wasn’t about new
technology, it was about a new application of existing technology.
Enter the role of challenge funds. What if a firm could use somebody else’s cap-
ital to overcome the internal competition (one hurdle down) and a compelling
proposition could be shaped that would give the company some comfort that the
project was addressing a market of potential future value?
And so it was in a conference hall in Johannesburg that I first gave serious con-
sideration to using a challenge fund to circumvent the constraints of our new
product development processes. I must make it clear, these corporate processes are
there for a very good reason—the management framework brings discipline to
project development and helps ensure that funds are spent wisely for the share-
holders. But equally it is also important to find ways to initiate new areas of busi-
ness that are higher risk; in this regard I believe that challenge funds have a strong
role to play.

66 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

A Public–Private Partnership Opportunity

In 2000, the U.K. government’s DFID established the Financial Deepening
Challenge Fund (FDCF), making available £15m for joint investments with the
private sector on projects that help improve access to financial services. Twenty-
eight projects have been funded in South Asia and Africa. The money was award-
ed on a matched basis (50% of total costs) and a competitive bid process.
Vodafone’s contribution could be in the form of people, budgeted at an agreed
rate. The FDCF fund managers and the proposal assessment team were looking for
innovation. This could involve the development of a product or service that was
not previously available in a target market, a new service that gave customers
access to goods or services that would previously have not been available, or the
application of a technology that reduced the costs of service provision. Many of the
successful applicants were large, well-known private sector companies that faced
challenges similar to Vodafone’s in pursuing what would perceived as low yield
projects.2
I spent a few weeks in mid-2003 putting together a proposal, whilst focusing
on two things: broad support for the concept from a couple of very senior execu-
tives in the company, and the buy-in of my colleagues on the ground in East Africa
(FDCF’s target zone). Both were fundamental to progress. If somebody at the top
of the company doesn’t provide executive sponsorship, things won’t happen and
the project would eventually be relegated to the scrap-heap of PowerPoint presen-
tations. And at a practical level, it quickly became clear that we faced a resource
issue on the ground. Safaricom is a very successful company, growing very quick-
ly, and its employees are all extremely busy growing the mobile customer base,
operating the network, and supporting customers with voice and text products.
Expecting the Safaricom team to engage in a complex project alongside their day
jobs would be a mistake. Instead, I decided to provide a dedicated project manag-
er who could lead all aspects of the development, integrating with the relevant
departments of Safaricom from IT, Operations, Customer Care and the senior
management team.
When I submitted the proposal, there were many unknowns. Much of the pro-
posal was weighted towards completing a preliminary needs assessment and not a
functional specification of a new product. The entrant of a telecom company into
a funding competition for the financial services sector took a few of the FDCF pro-
posal review team by surprise, but we overcame some initial cynicism and were
awarded funding of nearly £1 million, which was matched by Vodafone.
A few weeks later, contracts were signed and we set about organizing a series
of open workshops in Nairobi and Dar es Salaam. Invitees included banks, micro-
finance organizations, other technology services suppliers, nongovernmental
organizations with an interest in micro-credit, and representatives from the tele-
coms and finance sector regulators. We asked a question: Assume that the technol-
ogy can do anything you want it to; what are the biggest challenges you face in
growing your business or increasing access to financial services? We were able to

innovations / winter & spring 2007 67

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

identify two broad findings: (1) improved internal management processes would
allow organizations to speed up and improve the efficiency of delivery; and (2) it
should be very simple for customers to get access to finance.
The natural inclination of a service organization is to think about the con-
sumer first so we focused on number two. A pilot partnership was created between
us (the network operators), a micro-finance institute (MFI), and a commercial
bank. The view was that each
could bring to the project a dif-
ferent set of competencies,
If somebody at the top of the which would improve access to
company doesn’t provide finance for the un-banked
population. Network operators
executive sponsorship, things bring connectivity and a huge
reach through the airtime
won’t happen and the project reseller distribution network; a
would eventually be relegated to microfinance organization
understands the market need
the scrap-heap of PowerPoint for micro-loans and other
presentations. And at a practical financial services but is typical-
ly not a regulated bank; and a
level, it quickly became clear commercial bank brings the
that we faced a resource issue discipline and compliance
aspects of storing and manag-
on the ground. ing customers’ funds.
The proposition started to
firm up around the design and
test of a platform that would
allow a customer to receive and re-pay a small loan using his or her handset. We
wanted to allow the customer to make payments as conveniently and simply as
they do when they buy an airtime top-up, so a central feature of our proposition
was to use the distribution network of Safaricom airtime resellers (or Agents in M-
PESA terms) to facilitate this process. This service should also bring business effi-
ciencies for the MFI and allow it to grow its business more quickly and to more
remote locations than is possible using traditional paper processes.

STAGE 2. FROM TWO MONTHS TO TWO YEARS AND

LEARNING TO PUT CUSTOMERS AHEAD OF PROCESS
AUTHOR: SUSIE LONIE

Turning a Concept into a Pilot

Sitting in a comfortable office in England and deciding what Africa needs is an
approach doomed to failure. The market is littered with first-world solutions that
have utterly failed in emerging economies. For this project to be successful

68 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

Vodafone needed someone on the ground in Kenya, ensuring that the team prop-
erly understood the environment in which the service needed to work, and the
detailed product requirements. That’s where I came in. Because of my background
in Western-style mobile commerce, I was invited to accept a short secondment to
Kenya to complete a product specification and get things moving. In February
2005 I boarded a flight to Nairobi.
We had a fair idea of what we wanted the service to do, but were not sure how
to do it. The first big decision was buy or build? If we could buy software off the
shelf to meet our needs it would make sense to buy. So we went shopping and
found a multitude of financial service platforms with a fairly similar range of func-
tionality. Therein lay the problem: They had all been designed with Western bank-
ing infrastructure as the point of reference, and then added on other features. It
became clear that we would have to make some significant compromises around
the functionality and user experience if we bought one of the proprietary prod-
ucts. We reluctantly decided that we would have to bite the bullet and build our
own service from scratch.
This decision raises a key point which has reappeared regularly in every aspect
of this project. There is a fundamental difference between the way banks and tele-
coms operate. Mobile network operators are relatively young, entrepreneurial
companies that have experienced rapid growth and high profits through huge vol-
umes of low-value transactions. Banks tend to be mature organizations with well-
established business practices and a reassuringly cautious attitude to change. Their
business is based upon fewer transactions that generate relatively high margins.
The net result is that when a telecom decides to create a financial service such as
M-PESA, there is a collision of philosophies. So we decided to build.

And Now for Something Completely Different

The difficulty with doing something totally new is that you are not terribly sure
where you are going until you arrive. We knew a few things, but much of the time
we were guessing. This made the formal request for proposal (or RFP) something
of a working document, but there were some key principles to follow:
It was critical to understand the systems and capabilities of Safaricom, the local
mobile network operator that provides connectivity. Safaricom would also adminis-
ter the pilot on the ground, so gaining its commitment was a critical first step. An
understanding of Safaricom’s systems and capabilities had to underpin whatever
we developed, so that we could give it appropriate tools to run the commercial
service.
We were specifically targeting the unbanked. So whatever we designed would
need to operate in the absence of a consumer bank account. Therefore we needed
to hold whatever real money was in the system in a bank somewhere on the cus-
tomers’ behalf.
The e-money must always exactly match the real money or we could find ourselves
in the unfortunate situation of creating currency. We partnered with CBA

innovations / winter & spring 2007 69

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

(Commercial Bank of Africa) in Kenya to provide whatever conventional banking

services were required. The platform issues e-money to mirror real money in that
bank account.
The consumer interface would have to be a basic model mobile phone. One chal-
lenge was that smartphones are not abundant in Kenya. Thus, communications
channel options were limited to modes such as Voice and SMS (text messaging). It
took little time to determine that SMS was likely to offer the best compromise
between usability, security, and cost. Menu-driven access by SIM toolkit, which is
standard software on all SIM cards, was the obvious way to go.
We had to find customers with a real market need to use the service. This may
appear obvious, but the mobile commerce market is strewn with technical solu-
tions looking for a problem. We partnered with Faulu Kenya, a local MFI (micro-
finance institution) with several thousand borrowers, who typically run small
businesses. Most of Faulu’s customers repay a few dollars every week into the Faulu
bank account. The normal process for doing this is to form groups of about 20
people who meet each week and submit cash to the group treasurer. He in turn
takes the money to a local bank, accompanied by a suitable bodyguard of group
members. This may involve a long bus journey to the nearest bank. The weekly
loan repayment ritual is time-consuming, costly, and keeps people away from their
businesses. Here, we agreed, was a real market need where mobile commerce could
make a significant difference to people’s lives.
We needed retail outlets to act as M-PESA agents, where consumers could go to
deposit cash into or withdraw cash from their e-money accounts. Safaricom has a
large, established network of several hundred airtime dealer outlets across the
country where consumers buy prepaid airtime credit. Dealers are independent
companies with typically 5 to 20 retail outlets selling airtime, cellphones, and other
goods. So these dealers seemed to be an obvious initial source of agents if we could
figure out how to engage them. It was clear to us even then that we would need
other types of retail outlets in order to provide reach into rural communities, but
these dealers were a start.
We spent some time deliberating over whether agents should be given POS
(Point of Sale) devices and the customers given magnetic stripe cards. However,
visits to physical retail outlets (and thinking about scaling up a successful pilot to
a national launch) provided a severe dose of pragmatism. POS devices cost a lot of
money and need to be maintained. Perhaps some retailers in the city could afford
them, and perhaps we could convince them that the business opportunity justified
the investment, but there was not a hope that the poorer and more remote areas
could afford to participate. Instead, we opted for the rather elegant solution of giv-
ing the agent a mobile phone with a different M-PESA menu than consumers, cus-
tomized to their needs. The low-end phones we used cost about $40 each and
required no maintenance at all.
The RFP took account of all these points and many more; we tendered for an
external software developer; and chose Sagentia, a British company specializing in
“blue sky” strategic development. As we suspected, the RFP was far from exhaus-

70 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

SNAPSHOT: Mobile Money Speeds Commerce

Mary Mwangi has a small general store in Meru, Kenya. She has just received a
call from a family friend and occasional supplier of stock to her shop who lives
in Nyeri, nearly 100 miles away. She is told that for Ksh 7,000 (about $100 US)
she could secure a supply of kitchenware for less than half the usual price, as long
as she can pay for it today. This is a good opportunity for Mary as she knows that
she can sell these goods in Meru at a profit.
How does Mary secure this deal quickly? Time and distance are not on her
side. She doesn't hold a bank account and neither does the supplier. She does
have the cash but it is in her Meru store. She could send her money with a friend
on a bus to Nyeri but it will take most of the day and cost a significant part of
her profit. Traveling with money is also a risk as highway robbery is not uncom-
mon.
The answer lies in M-PESA (pesa is the Swahili word for cash). Mary recent-
ly registered with Safaricom to open an M-PESA account. This was a simple
process that gave her access to an e-money account managed entirely through
some simple menu instructions on her prepaid cellphone. Ten minutes after the
call from her friend, Mary has been to a local Safaricom Airtime Dealer (of
which there are several in Meru) and has deposited Ksh7,000 into her M-PESA
account. This is very similar to topping up he prepaid cellphone airtime, except
she is loading cash into her M-PESA virtual account. A few minutes later Mary
has returned to her shop where she sends an SMS text message instructing M-
PESA to transfer half the cost of the goods to her friend's M-PESA account,
effectively securing the purchase with a real time funds transfer. The goods are
dispatched to Meru on the next bus, and when they arrive Mary settles the
remaining money by sending another text message instruction to the M-PESA
service.
Making this payment quickly and securely by cellphone cost Mary Ksh60
(less than a dollar).
The M-PESA service is fast, secure, and very cost-effective. It is opening up
new opportunities for businesses like Mary's all over Kenya as well as support-
ing person-to-person money transfers, or remittances, which are common in
many economies where the bread winner supports an extended family, often
many miles away.

tive; it was important that Sagentia be extremely open-minded and flexible, create
a very configurable system, and show a strong willingness to get involved in defin-
ing the detailed functionality. In fact, Sagentia demonstrated the required skill set
and attitude many times during this project.

innovations / winter & spring 2007 71

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

Where Angels Fear to Tread

The list of tasks to get the pilot working was prodigious. Our total human
resources amounted to Nick in the U.K. with me on the ground in Kenya sorting
out the practicalities full-time; as much of Sagentia’s time as we could afford; and
any help we could coax out of whomever came our way, especially from a number
of Safaricom employees. We were fortunate that the project inherently appealed to
people. It is unusual in a large, competitive commercial business to work on a proj-
ect that is truly ground-breaking but small enough that each individual can make
a big difference, and to address a need which could materially change the lives of
many people for the better. So help was forthcoming, but at a cost: Each member
of the core team put in effort significantly beyond the call of their contracts.
The first thing we had to sort out was where to locate the servers. Our initial
assumption that they would need to be housed in the local bank hosting centre was
incorrect. The pilot service was principally supporting microfinance transactions,
which is currently unregulated in Kenya. We were advised that if a bank hosted the
service it would lead to unnecessary complications and delay. We arranged for the
servers to be placed at Safaricom instead. They eventually arrived after an extend-
ed delay due to incorrectly copied paperwork in the Middle East and various cus-
toms issues. Once installed, the U.K.-based technical team gamely battled to work
with them. Internet speeds in Africa are often not quite as fast as in the West, and
after several weeks of waiting over a minute between clicks from one transaction
to another it was clear that we needed to rethink matters before the techies
rebelled. So the main servers were relocated to the U.K. and connected to Kenya by
an internet protocol (IP) link. The developers then happily got on with tapping in
code.
The service needed a name. A local advertising agency quickly suggested M-
PESA. You may recall that pesa means money; dial M for mobile, and a short, sim-
ple, catchy brand name is born.
We needed a source of Faulu clients who could participate in a trial of M-
PESA. As this was a pilot and we expected that there would be various issues, it
made sense to start with groups in Nairobi who were accessible. Two locations were
identified: the City Centre and Mathari, a slum about a 20 minute drive from the
centre of town. The pleasant market town of Thika, about an hour up country, was
confirmed as the third location for the pilot. Faulu then recommended a few
groups most likely to understand how to use a cellphone and embrace the service
so we could get off to a flying start. Other (less mobile-phone literate) groups were
deferred for later.
Meanwhile we needed to finalize the consumer handset experience based ini-
tially on English text. Whilst many Kenyans speak pretty good English, the further
from Nairobi you go, the more Swahili and tribal languages take over. As one of the
key requirements of our funding was to extend financial services to rural areas, we
decided early that we needed to be dual language.3 All the phone menus and SMS
responses needed to be translated. One a challenge was getting the Swahili mes-

72 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

sages down to no more than160 characters. By contrast, English is a very compact

language.
It was unexpectedly straightforward recruiting agents where customers could
load cash in or take cash out for the pilot. One sales team identified the most
appropriate existing airtime dealer stores in the areas where the pilot customer
groups met, and we paid the head offices a few visits to explain the service and the
potential benefits for their businesses and for Kenyans generally, and they quickly
agreed to become M-PESA agents. The underlying cause of their ready support
was the remarkable success of mobile phones in Kenya; five years earlier Safaricom
had a few thousand subscribers, and by the pilot that number had risen to 5 mil-
lion. The dealer businesses had been built upon mobile operator innovations, so
for them this was simply the next innovation.
One small hitch was the absence of potential agents in Mathari. But such is the
entrepreneurial nature of Kenyan business people that one of the agents saw this
as an interesting gap in the market. She paid a visit to the area, fell into conversa-
tion with a local petrol station owner, and a few weeks later had built a small shop
on his forecourt that became a thriving business in Safaricom airtime, as well as
supporting the M-PESA pilot.
Within Safaricom, there were two key groups of people involved in the pilot.
The first was Customer Service, to take calls from consumers or agents with issues.
A dedicated M-PESA line, 234, was created for their use. The second was Finance,
to manage the flow of cash to and from the M-PESA bank account and accurately
reflect it in the M-PESA system.
Internal training sessions were organized to show how to use the handsets and
screens, and to underscore the basic principles by which the service worked.
Operating instructions, frequently asked questions, and escalation procedures (an
incremental process for managing problems) were created. Ongoing coaching on
this new type of business was used to ensure everyone knew what they were doing.
In addition, occasional resources were needed from Engineering, Billing, Risk,
Operations, Sales, IT, Marketing, Products & Services, and just about every other
department. Beyond a general instruction to the business from on high to be coop-
erative, staff had absolutely no direction in their job descriptions to give the level
of help that they did. It says much for the nature of both the people and the proj-
ect that they rallied to the cause and were incredibly supportive.
Coming to grips with Faulu’s back-office requirements was more difficult.
Faulu uses a paper-based system that involves collecting hand-written records
from the field every week, and having a room full of people input the data in its
central microbanker system. Faulu had difficulties in managing the manual proce-
dure and the M-PESA pilot simultaneously and so decided that their best route
forward was to eschew the benefits of an automatic real time data entry system and
replicate each of their manual processes electronically. This included replicating
the group payment structure by cellphone and having the same data input clerks
print out and manually input all the electronic M-PESA records.
As the time approached to begin the pilot, Faulu became understandably cau-

innovations / winter & spring 2007 73

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

tious about the robustness of the system and the security of its money. We had to
run a familiarization trial with Faulu’s internal staff for several weeks to give them
greater confidence in and familiarity with the system. It slowed us down but
allowed us to practice our training techniques.

Lift-off
On 11 October, 2005, the pilot finally started. Eight agent stores were given their
M-PESA phones and repeated training sessions. Nearly 500 clients were enrolled,
given phones, and instructed to use M-PESA to repay their loans. Their incentive
was a free phone and a few dollars in their M-PESA accounts.
The first obstacle we encountered was the agents’ hesitation to pay out cash
withdrawals. No matter what the training told them, it was a brave shop assistant
who opened the employer’s till and handed out cash because they had been sent a
text message telling them to do so. We overcame this problem by giving agents sep-
arate M-PESA cash floats, along with reassurances from their head offices. I also
visited agents every day to help with the cash withdrawals until they got used to it.
Consumer training was quickly identified as being probably the biggest chal-
lenge. The challenges included:
Familiarity with phones: There was a great divide between people who were
familiar with mobile phones, and people who were not. The former tended to pick
up M-PESA quickly. For the latter group, the first chunk of any training session
was taken up by explaining the concept of a menu, showing them how to find M-
PESA, how to find their SMS inbox, etc.
Training environment: It’s no surprise that training smaller groups in comfort-
able surroundings definitely work best. I remember one nightmare session in a tin
shed with the mid-day sun beating down, an enthusiastic 40-a-side soccer match
going on outside, and about as many potential clients wrestling for hours to learn
how to use M-PESA. It says much for their persistence that they learned the sys-
tem.
Task complexity: The pilot involved quite a number of consumer transactions.
Because of the need to follow Faulu’s paper processes for loan repayment, clients
paid a treasurer group account; once the treasurer had all the SMS payment
records, he paid Faulu by SMS. In addition, clients could send money person-to-
person (P2P), deposit cash in or take cash out with an agent, and also needed to
contend with account administration, such as changing PINs. Cramming all of this
into one session was a tall order.
It was clear that we would need to find a way to simplify things before launch-
ing a national service for millions of customers.
A month or two into the pilot we decided that we could expand the volume of
business by allowing consumers to buy prepaid airtime with their M-PESA e-
money. Thus we created a new menu item and connected to the Safaricom billing
system. More consumer training ensued, but as a hook we offered an introducto-
ry 5% discount on any airtime bought by M-PESA for a few weeks. The results

74 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

Figure 2. Overview of the M-PESA service.

innovations / winter & spring 2007 75

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

were gratifying. We even saw a number of customers setting up as informal airtime

resellers as a side business.
There were several instances of this kind of entrepreneurial behavior. We
watched the transaction patterns closely on the web screens and speculated upon
what was going on. When we could stand no more we organized researchers to
explain some of the transaction patterns. Aside from the standard loan repayments
for which we had designed the system, we observed several other applications:
• People repaying the loans of others in return for services;
• Payment for trading between businesses;
• Some of the larger businesses using M-PESA as an overnight safe because the
banks closed before the agent shops;
• People journeying between the pilot areas, depositing cash at one end, and
withdrawing it a few hours later at the other;
• People sending airtime purchased by M-PESA directly to their relations up
country as a kind of informal remittance;
• People outside the pilot population being sent money for various ad hoc rea-
sons; for example, one lady’s husband had been robbed, so she sent him M-
PESA to pay for his bus fare home;
• People repaying loans in return for cash on behalf of a few colleagues who had-
n’t mastered the use of the phone—or simply sold it .
Initially we had problems with clients losing their SIM cards—not the phones,
just the SIM cards.4 How could this be? We learned that perhaps half of our clients
already had phones and rather than carry both phones around, they tended to
carry the M-PESA SIM in their wallet and swap the SIMs when they wanted to do
a transaction. SIM cards are small, easy to drop, and easy to lose when not wrapped
in a phone. It also became apparent that if M-PESA was not readily available on
the SIM, the number of spontaneous transactions was going to suffer. After issu-
ing our first dozen replacements it became clear that we needed to fix this and
thought of SIMEX cards. SIMEX are SIM cards without associated phone num-
bers—if you lose your phone but want to keep your number when you get a
replacement, the SIM used is a SIMEX with your old number transferred to it. So
we ordered some M-PESA SIMEX, transferred the customer numbers to them and
moved their M-PESA accounts to their own phone numbers. The number of lost
SIMs dropped to negligible levels, and the number of transactions increased. It
seems obvious now, but it was a big step forward for the success of the pilot.
It was clear that the vast majority of group members strongly appreciated the
convenience and security benefits from using M-PESA for loan repayments. This
was not entirely good news for Faulu; clients no longer had the same compelling
need to attend the weekly group meetings. Opinion is divided upon the value of
MFI group meetings generally; Faulu was strongly in the supporters’ camp and
consequently found the drop off in attendance disturbing.
Over the duration of the pilot it became apparent that the service was more
user-friendly for the clients than for the microfinance institution. Kenya Faulu had

76 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

ongoing difficulties with its internet connection, even after we provided a dedicat-
ed mast and satellite link. Reconciliation of the M-PESA and conventional
accounts was a headache, and loan disbursement required significant over-riding
of Faulu’s existing automated processes. The bottleneck in transferring the money
M-PESA had collected in loan repayments to Faulu’s bank account was getting its
books to a point where it could request the funds.
To make M-PESA more suitable for institutions like MFIs, we need to create
data export files that can be easily uploaded in whatever format their existing soft-
ware requires. Ideally, we should create an interface directly between their systems
and M-PESA. This is not a difficult task—M-PESA keeps detailed records of every-
thing; we just need to know what data is required and in what format.
Meanwhile, the benefits to consumers of using M-PESA to help manage their
personal finances were obvious and compelling.

Turning a Pilot into a Launch

By the end of the pilot, executives at both Safaricom and Vodafone realized we
might be onto something big, despite M-PESA’s unusual origins. For Safaricom,
the big opportunity was to extend its service into a completely new kind of busi-
ness as a Payment Service Provider with a new revenue stream, and to use this to
increase customer retention. M-PESA charges a transaction fee to the sender of
funds. These charges vary by the type of transaction and the amount of funds
being moved. The service itself is free and all deposits made into the M-PESA
account are free. There is no impact on the customers’ airtime’ credit for any M-
PESA transactions. (Current tariffs can be viewed on <www.safaricom.co.ke>.)
For Vodafone, it became clear that the M-PESA system formed the basis of a
low-cost International Remittance Service. Globally, international remittances are
a $300 billion business, fuelled by migrant workers sending money home. Many of
the busiest remittance trails, such as Germany to Turkey and Europe to India, are
in territories where Vodafone has a presence. So getting approval to launch in
Kenya was pretty straightforward. Doing it was, of course, another matter.
The pilot officially ended on 1 May, 2006. In practice it ran on until October
since nobody really wanted to stop using it, but the data collected through May
gave us the information we needed to make decisions. We had evidence that the
product functionality was right, both technically and commercially, to provide a
service that could deliver a healthy profit and meet consumer needs, both in Kenya
and other developing economies. The next step was to get agreement from
Safaricom to support a Kenyan launch, which meant preparing a detailed launch
product definition. The pilot experience had shown that a mass market launch
with an MFI would be too complex, so we needed to find a compelling consumer
proposition around which everything would be based.
A workshop with the Safaricom commercial team identified the key proposi-
tion as Send Money Home. In Kenya, as in most developing markets, many fami-
lies have a small number of breadwinners who work away from home in order to

innovations / winter & spring 2007 77

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

provide for the others. Practically every Kenyan I worked with sends money up
country to some family members. They use various means to do so, ranging from
sending heavily disguised parcels by bus or finding someone travelling that way
and giving them the cash, both of which are risky in a country where highway rob-
bery is literally a commonplace event, to more formal mechanisms such as Western
Union and the like, which are less common as they are very expensive and there are
few cash outlets in rural areas.
The launch service was therefore limited to three features providing the rela-
tively simple functionality which made the consumer proposition much easier to
understand and to use. Users could deposit in or withdraw cash at agent stores;
transfer money person-to-person (P2P) ;and buy prepaid airtime.
However, every other modification we made to the service added complexity.
In the latter stages of the pilot we had hired consultants from the electronic bank-
ing world to audit the system and make recommendations. Their suggestions
included a major change in the design of the agent phone menus. In the pilot, each
agent SIM was associated with one e-money float and one user. In practice, sever-
al shop assistants used the same phone and shared the not-so-secret PIN. A funda-
mental change was to split that one agent phone entity into four or more separate
entities associated with the e-money float: the store; the M-PESA till; a primary
assistant responsible for administering the till; and one or more standard assistants
who could also use that till to do M-PESA transactions. Each operator is now iden-
tified by a unique Agent ID, usually their initials, and a genuinely secret PIN.
An ongoing issue was that agents ran out of e-money float. When customers
deposited more cash than was deposited at an outlet, the e-money float was
reduced, sometimes to the point where the agent ran out. To replenish this float the
agent had to return cash to the M-PESA bank account so it could be reflected as an
increase in M-PESA float. This could take several days to clear during which time
the agent could not properly serve customers. For the launch, however, each agent
would have several stores with e-money float, not just the one pilot outlet. This
gave us the idea to create a whole new function we call float balancing. The logic
goes that city stores will see lots of deposits from people wanting to send money
home, and rural stores will see more withdrawals as people cash the money they
have been sent. Whilst the money will never balance exactly in one agency, if we
could provide a tool for agents to move float from stores with excess e-money to
those that were running out, this could help relieve the problem. A web tool was
duly created for the agent’s corporate accountant to do this. For Head Offices with-
out internet access, and for times when it is unavailable, we also created Head
Office handset menus to do a similar job.
These and a wealth of other technical changes were made to the system. We
also had to tackle any number of administrative requirements. Safaricom and
Vodafone have formal, structured procedures that any new service introduction
must follow, but for the pilot we had blithely ignored all but the most pressing of
these with a view to getting the job done without drowning in bureaucracy. This
was marginally acceptable for a small pilot, but for a full commercial launch we

78 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

had to resume playing by the rules. It took about four months to get the paperwork
sufficiently buttoned down for a corporate launch.
Neither Safaricom nor Vodafone has a banking license. This means that man-
agement of the legal and regulatory structure of the business was a delicate matter.
Many rounds of discussions with Kenyan and English lawyers, many straw men,
and many heated debates later, we came out with a complex legal structure appro-
priate to running the M-
PESA service in Kenya—
operated locally by It is clear that regulation of
Safaricom, but owned, host-
ed, and developed by services such as M-PESA will
Vodafone. A new trust com- happen sooner rather than later.
pany was created.
We then needed to cre- This is no bad thing for either
ate a department within
Safaricom to launch and
consumers or service providers as
run M-PESA . The first step long as the regulation protects the
was to recruit someone to
head the department.
consumer against the risks
During the course of the involved. The better the regulator
pilot I had identified the
ideal internal candidate understands the capabilities and
who was heading another limitations of services like M-
department at the time. It
took a while to convince her PESA, the better and more
to accept the role, but since appropriate the regulation will be.
she agreed she has thrown
herself into it with all the
competence and enthusi-
asm expected. Next came recruiting the rest of the team from Safaricom employ-
ees and external candidates. This was a challenge as there was no existing pool of
subject matter experts. Therefore, the most important criteria were a good educa-
tion, experience in telecoms or banking, and most important, the right attitude.
We busily prepared training materials for the agent stores in both English and
Swahili, but the more we considered the logistics of providing training to 600 assis-
tants scattered around the country before launch, the less likely it seemed that this
could be adequately managed by our training team of four. The pilot had taught
us of the need for repeated training sessions and repeated store visits post launch.
Kenya is a big country with small roads, and to drive between towns is not easy. So
we added a new line to the budget to pay for a sales training agency. Now, fifty
trainers visit our agents on a regular basis to give any support required, provide
merchandise, and generally keep an eye on things.
The Central Bank of Kenya clearly needed to be engaged regarding financial
service regulation. We had met with the bank a few times during the pilot,

innovations / winter & spring 2007 79

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024

 Nick Hughes and Susie Lonie

although the numbers and consequent small risk made it of little interest to them.
When we approached the Central Bank regarding the national launch, it was quite
another matter. There followed a series of product demonstrations, requests for
documentation, compilation of information, more questions, meetings of clarifi-
cation, submission of a formal legal opinion, and so forth. E-money products such
as M-PESA are new to Kenya so there is no clear regulation yet in place.
Nevertheless, it was impressive how quickly the bankers’ questions progressed
from fairly basic to insightful and quite tricky. But we had done our homework
and eventually the bank confirmed that it had no objection to the service launch-
ing. Ten days after receiving this letter, we launched.
It is clear that regulation of services such as M-PESA will happen sooner rather
than later. This is no bad thing for either consumers or service providers as long as
the regulation protects the consumer against the risks involved. The better the reg-
ulator understands the capabilities and limitations of services like M-PESA, the
better and more appropriate the regulation will be. It is our intention to work with
the Central Bank to provide the information required to make informed decisions
as formal controls are introduced.

Reflections and Ramifications

The excellent early adoption rate of M-PESA in Kenya strongly suggests that the
service meets a need in the market. Usage is significantly above expectations.
Vodafone is already piloting the product in new markets will soon allow person-
to-person transfers across international borders. Vodafone has also created a cen-
tral team to coordinate the growth of this business.
If we agree with the assumption made in the Millennium Development
Goals—that the private sector must be involved in the development agenda to
achieve truly sustainable results—then challenge funds provide a useful mecha-
nism to facilitate private-public sector partnerships. Through DFID’s matching
challenge grant, corporations such as ours have been able to reduce internal com-
petition for capital, thereby allowing socially beneficial projects that might hold
higher risk or have lower returns on investment to go ahead.
This experience also has reinforced the insight that there is no substitute for
spending a significant amount of time at the start of a project on the ground
assessing customer’s needs well ahead of designing the functional specification of
any technology-based solution. We also learned to keep it simple. When it came to
moving from pilot to live system, a significant amount of the complexity in the
product was stripped, allowing Safaricom to go to market with a very simple con-
sumer proposition. Training is still a priority, but the burden is somewhat lighter
than in the pilot. The disadvantage of putting consumers first was that the busi-
ness requirements, such as Faulu’s with back office information management sys-
tems, proved to be an obstacle to scale. It is clear that we need to provide a solu-
tion that allows easy integration with financial institutions.
A partnership approach was at the centre of the project. All parties had to be

80 innovations / winter & spring 2007

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024
 M-PESA: Mobile Money

prepared to embrace the consequences of change –and this is not easy for natural-
ly conservative organizations that are not accustomed to working together. For this
and all the other challenges we described, getting senior sponsorship along with a
committed project team was critical.
We have since identified a plethora of new ways M-PESA can be used in Kenya
and elsewhere. There is market demand for faster, cheaper, and more efficient ways
of moving money for many reasons, ranging from bill payments to salary pay-
ments; affordable social payments to extending existing financial services; conven-
ient-low cost international remittances; and many more. These have been priori-
tized and development is underway. M-PESA has just taken the important first
step in what we hope will be a long journey in bringing people into the communi-
cations and financial mainstream worldwide.

1. The Millennium Development Goals are a set of eight goals established through the U.N. with the
aim of halving global poverty by the year 2015. For more information, see:
<http://www.un.org/millenniumgoals/>
2. Full details can be found at <http://www.financialdeepening.com>
3. Swahili is a hybrid language consisting of Bantu, Arabic and a smattering of English which is the
means by which the 42 tribes in Kenya communicate. Effectively it is a successful East African ver-
sion of Esperanto.
4. Subscriber Identity Modules (SIM) are removable cards that allow users to change phones easily
by removing the SIM card and inserting it into another mobile phone, thereby eliminating the
need for activation of the new mobile phone on the network. Wikipedia:
http://en.wikipedia.org/wiki/Subscriber_Identity_Module

innovations / winter & spring 2007 81

Downloaded from http://direct.mit.edu/itgg/article-pdf/2/1-2/63/704171/itgg.2007.2.1-2.63.pdf by guest on 15 September 2024