Nathan Wyeth

Report from the Mobile Money Summit (Part 3) – Learning from Agent Networks

This is part three of a three part report on the Mobile Money Summit. You can read part one here and part two here.

The mobile revolution in places that have previously had minimal modern communications infrastructure was not simply a process of strong cash flow from airtime usage paying for building cell towers and related infrastructure. Extensive networks of tens of thousands of outlets for purchasing pre-paid airtime, not to mention purchasing and repairing phones, have been created virtually overnight.

Now that cell towers have been built and airtime distribution channels established (essentially product distribution) every next iteration of the mobile communications revolution, and the platforms built on top of it, are going to get more complex.

Mobile network operators are now engaged in building new networks on top of their existing distribution channels – agents trained and equipped to deliver mobile money services. This process is not only more complex than selling airtime, it’s unfolding before our eyes and was a major topic at the Mobile Money Summit.

The distribution channels for products like Coca-Cola are cited frequently as examples to learn from for deveopment purposes. But the more transformative a new product or service is for a base of the pyramid community, the more complex it is likely to be as compared to carbonated water.

These agent networks are, in contrast, a work in progress but an example that can be learned from for the distribution of products and services that are new, highly complex, and require extensive trust and education.

The agent network for a mobile money service is an essential offline side to services otherwise delivered electronically. If mobile money is to reach people for whom traditional financial services and electronic money are not already accessible, especially where this is a question of geography, it is a mobile operator’s agent network that will change this.

Furthermore, the value of a mobile money service, especially for person-to-person transfers but also for any number of transactions that are most arduous using cash, is defined precisely by how remotely the agent network reaches – if it’s only present in places that are already easily accessible, it’s probably relatively easy to move money there already.

The Mobile Money for the Unbanked program of the GSMA, funded by the Gates Foundation, has focused on the development of agent networks as a key factor in achieving scale for mobile money services and for ensuring that actual benefit is delivered to the poor. At the end of the day, as one panelist said at the Mobile Money Summit, if someone walks all day to make or receive a mobile payment and the agent they reach has no cash and can’t help them, they’re not going to come back.

Some of the MMU’s findings on what works for building these networks is available on its website.

  • As would be expected, some basic business imperatives apply to agent networks. At the most basic level, if being an agent isn’t profitable, it won’t matter how good a service is provided to customers, there will be no agents to deliver it. But attractiveness is not just a question of individual transactions, but customer/agent ratios, the speed with which agents get paid, and relative profitability compared to selling airtime or other goods.
  • Kenya’s Safaricom, which has been studied more carefully than others because of its breakout mobile money service M-PESA, has had to carefully manage growth in its agent network in tandem with the growth in its customer base in order to keep agents with enough customer transaction volume to stay involved. The M-PESA customer-agent ratio is currently close to 500, down from 1000 when there were fewer agents.
  • Inventory management is something generally left to retail businesses by their distributors. But mobile money is a much more ’real-time’ distribution platform, in which the fate of an operators system is very much linked to whether their agents have cash on hand whenever a customer wants to cash out. Managing the ’float’ that remote small businesses have to deal with adds a significant layer of complexity to operating a mobile money platform.
  • Not only do agents need to have a good understanding of the ins and outs of the particular mobile money platform they’re interacting with, they need to convey this to customers accurately. A high level of trust is needed before people will start handing over their cash to an agent, and especially before leaving their money on an e-wallet when they receive it, which only exacerbates the need for cash on hand to service cash-out requests.

For more on the challenges and lessons from building agent networks, you can check out presentations from the GSMA’s MMU Working Group as well as CGAP here, here, and here (all pdf).

Despite the challenges, it’s important to remember how incredible it is that these highly responsive, complex netwoks are being built, and the lengths to which operators are going to deliver these services. Even in remote parts of Afghanistan, Roshan’s network for its M-PAISA service is setting up agents with enough cash to do payroll for Afghan security forces.

When police don’t get paid is when their allegiance to the government comes into question – and can be tempted by the opium money that the Taliban can offer.

Many of the interactions with mobile money platforms may be more mundane than this from an international security perspective. But to deliver new products and services to the poor will likely involve either interacting directly with these emerging pillars of the networked economy in developing markets, or learning extensively from their success.