M-Pesa: a Mobile Money success story from Kenya
M-Pesa is a phone-based peer to peer money transfer service launched by Safaricom, Kenya’s largest telecom provider, in 2007. According to a Kenyan government report published from 2012, just five years after launch, there were 19.5 million mobile money users in Kenya (representing 83% of Kenya’s adult population), transferring ~US $8 billion per year (~24% of Kenyan GDP).
Banks and telecoms around the world have initiated hundreds of mobile-money deployments in the past six years attempting to replicate M-Pesa’s widespread adoption, but with limited success. Only a handful of these initiatives have even reached sustainable scale.
Although some of M-Pesa’s initial success could be attributed to a uniquely favorable context for mobile-payments (strong customer need, welcoming regulatory environment, support from banks, strong brand awareness of Safaricom), its rapid and sustained growth was only possible due to a thoughtful operating model design, particularly regarding M-Pesa’s “agent network.”
Let’s have a look at what M-Pesa is, how it works, and the key operating model decisions Safaricom made to ensure M-Pesa’s long-term success.
What is M-Pesa and how does it work:
Customers who sign up for the M-Pesa service can convert between e-cash and real cash (these are called cash-in / cash-out transactions), and can transfer e-cash from their account to that of another account holder via SMS.
Cash-in / cash-out operations take place at one of many designated M-Pesa retail outlets, also known as “agents”. These agents are not employed by Safaricom, but are simply retailers / regular businessmen and women that are ‘authorized’ to trade e-cash for real cash.
Business Model
M-Pesa creates economic value for Safaricom in two primary ways:
- Revenue from transaction fees that Safaricom collects via the agent during cash withdrawal operations and transfer operations (depositing money into mobile wallet is free).
- Reduce Safaricom customers’ churn, improve engagement, lifetime value etc.
Safaricom pays commission to its “agents”, usually on a monthly basis, based on metrics such as transactions per branch, customers per branch, and quantities transacted, etc.
Because it takes agents a couple months to ‘ramp up’ at their branch by attracting M-Pesa customers and convincing them to start transacting, the business model of M-Pesa incurs significant up-front costs and is one of the reasons many mobile-money deployments fail in the early days.
Mobile-Money becomes profitable only when it goes viral. According to a McKinsey report, to make mobile money for the unbanked commercially viable, operators and telco’s like Safaricom “must sign up 15 to 20 percent of the addressable market.”
Operating Model and Key Decisions
1. Grew agent network same pace as customer base
Given the up-front costs of acquiring agents, it is tempting for mobile money providers to want to take short cuts and minimize the agent-to-customer ratio. However, this does not set an individual agent up for success.
If Safaricom were to recruit too few agents, customers would find M-Pesa difficult to use and difficult to access.. On the other hand, if there were too many agents, many of them would not be able to generate enough business to cover the cost of managing their e-cash and cash liquidity. As a result, they would stop maintaining their electronic money float and cash balances.
M-Pesa’s success lies in the fact that they grew their agent network at the same pace as their customer base, keeping transactions per agent per month steady at around 1,000 / agent / month.
2. Invest time and money in recruiting agents with the right skills and incentives.
Safaricom realized early on that they could not use their pre-existing airtime dealers as agents, since the airtime dealers were disincentivized to sell mobile-money. Mobile money would cannibalize offline airtime purchases since consumers would start topping up their phone airtime online / via the mobile money app instead of showing up to the Agent’s stall.
So, they set out to create their own Agent network. They talked to mom-and-pop stores and retailers to determine level of education, conversation ability, business acumen, and once they recruited an agent, they offered training on customer service, how to manage liquidity with cash and e-cash, etc. Even post-hiring, Safaricom continued to monitor agents through site visits once every two weeks at each branch.
3. Geographically spread out agents to ensure proximity to consumers.
According to a McKinsey report on Mobile Money, proximity of nearest agent makes a significant impact on transaction volumes. “When a cash agent is more than 15 minutes away, mobile money has relatively little appeal, and customers use it once or twice a month. But when the agent is less than 10 minutes away, usage rises to 10 times a month—and for those within 2 minutes of an agent, to 30 times a month.”
Safaricom spread its agents out across Kenya so as to truly enable network effects and enable Kenyans to send e-cash to their family members and friends even if they did not live in the same geography.
4. Hub-and-spoke for Agents and Super-Agents
By 2012, Safaricom had over 40,000 agents. Rather than interacting directly with each of these agents, the company selected some agents to be “super-agents” which would have higher e-wallet sizes and therefore would buy greater e-cash from Safaricom up-front and then sell this e-cash to other sub-agents in their geography. This way, Safaricom kept its operations lean and only had to transact with the few superagents, who would in turn transact with sub-agents who would transact with the end-customers.
The super-agents became the hubs, the sub-agents became the spokes, ensuring ‘last-mile delivery’ of e-cash services to M-Pesa end-customers.
References:
http://www.mckinsey.com/insights/social_sector/mobile_money_getting_to_scale_in_emerging_markets
http://www.economist.com/node/16319635
http://blogs.ft.com/beyond-brics/2012/05/28/mobile-money-kenya-good-india-bad/
http://www.economist.com/blogs/economist-explains/2013/05/economist-explains-18
Great to see how local, tailored initiatives can introduce and scale innovations. I have two questions on this:
1/ Do we know if the Kenyan Government provided any support towards this?
2/ What’s been their strategy and results post 2012? I can imagine, given the success of this, there would have been other competitors (with deep pockets) who would have tried to grab share from m-Pesa in recent years.
Great article. A few thoughts:
1. One other way Safaricom makes money is the float on the money brought into the system. They charge relatively higher fees to transact outside Safaricom’s network.
And Ameya, to your questions.
1. The Kenyan government is a majority shareholder and many people believe that’s why it’s been successful. There was minimal regulation and it grew organically, a situation difficult to replicate in other emerging markets. Possibly why MM hasn’t really taken off outside Kenya
2. Their strategy is to raise barriers to entry. They reinforce their network effects by charging higher fees to transact outside the M-Pesa network. The also improve stickiness by several value-added services which their scale allows them to do. That said, there are several competitors: Come to the African Business Conference at HBS to learn more! Feb 26-28th 2016 – http://www.africanbusinessconference.com
Interesting business model and article!
The agent-focused model you describe brings up a few questions for me. If Safaricom is only dealing with the first layer of agents, who then in turn work with the subagents who work with customers (and I imagine in many cases there is even another layer or two), how does Safaricom try to control the interactions with customers? It sounds like they used to do site visits, but you mention they now operate (understandably) at arms length. M-Pesa is so linked to the Safaricom brand, which as you mention is very strong, but I imagine Safaricom has to work hard to flow training, brand messaging, key updates etc through its dispersed distribution network. They have the same challenge with their airtime distribution network, of course, which brings me to my second question – is there still little or no overlap between those two agent networks? I understand the initial concern of airtime sellers losing their business, but given the likely frequency of cash in/cash out transactions it seems you could make the case for investing in this new type of business to replace any lost airtime sales. I wonder if they have evolved into combining those two forces or if they remain focused on keeping them separate. Finally, I think the thesis of your post is that M-Pesa’s success is largely thanks to its agent model and footprint, or at least that these were necessary contributors to its ability to grow. Do you think this is why others have failed where M-Pesa succeeded? Or do you think others have understood that element but face different regulatory/environmental challenges?
TOMChallenge, to your question:
This piece reminds me a lot of the Facebook case and our thoughts on why Facebook was able to succeed where other social networks failed. In both cases, the company grew at the right speed, allowing them to scale their systems and infrastructure (agent network, working capital needs, and computer infrastructure in this case, only the latter in Facebook’s case) in a way that allowed for sustainable growth. Also similarly, the focus on the geography of the agents reminds me of how Facebook was able to increase their network effect and the marginal benefit of each incremental user by expanding to a few campuses at a time. I think a big piece of M-Pesa’s success has been this network effect. In the beginning, they built it effectively as they scaled, but now, similar to what we see with Facebook today, the scale that has been achieved and the associated power of the network effect serves as a barrier to entry to supplement any such barriers put in place by the Kenyan govt. In M-Pesa’s case, however, they are able to make the barrier even stronger by charging a higher price for transactions outside of the network (which is not a lever that Facebook had).