By: Paul Leishman: August 20th, 2010
By now, most mobile money practitioners recognize that for a service to succeed, the agents that support it need to be making money. In recent months, weve written about how mobile network operators have carefully managed their customer/agent ratios over time to ensure agents are provided with a sufficient volume of transactions to stay interested but volume is only half of the equation. To keep agents happy, mobile network operators must also ensure the commission models they present to agents are well designed.
In this post, Ill describe four pitfalls that MNOs should avoid when designing their mobile money agent commission models:
Fixed percentage based commission models are notoriously tricky to implement. In some cases mobile network operators are drawn to them on the basis of their simplicity; agents need not memorize a complicated guide instead they just remember that they earn X% of transactions they process, regardless of value. Unfortunately, MNOs that have implemented percentage-based commission models often highlight the reluctance they encounter from agents when it comes to low value transactions: that is, agents are happy to earn 2% from a $50 transaction ($1 revenue), but less enthused to spend valuable time earning 2% from a $9 transaction ($.18 revenue). This can lead agents to provide poor customer service (i.e., refusal to serve) when users seek to perform low-value transactions.
Models that offer agents a fixed commission for each transaction, regardless of value, are tricky to implement too. But in this case, problems arise from high-value rather than low-value transactions. To bring this challenge to life, lets consider the case of a rural agent in Laos. Each day this agent travels by bus to a bank branch and loads her e-wallet with $1,000 of e-money by depositing $1,000 cash. Back at her store, she has an opportunity to earn $0.25 for each cash-in transaction that she processes, regardless of its value. On a profitable day, this agent might serve 20 customers, each depositing $50, and earn $5. Her nightmare day, on the other hand, would be one in which a single customer asks to deposit $1,000 the entire value of her e-money stock and limit her days revenue to just $0.25 (not to mention the costs she incurred travelling to the bank to load her e-wallet). Here again, its easy to imagine the agent refusing to take the deposit, preferring to perform lots of smaller transactions over the course of the day. In practice, there are likely very few instances of customers asking to deposit the massive amount described above, but this scenario does reveal an important principle: when fixed value commissions are at play, agents have no incentive to process high-value transactions.
In some cases, mobile network operators have designed models where agents are offered lucrative commissions for facilitating cash-out transactions, but small commissions (or even none at all) for cash-in. For reasons that have already been well documented, this approach is flawed: as Eijkman, Kendall, and Mas wrote in our 2010 annual report , rural agents are likely to predominantly perform cash-out transactions, whereas urban agents are more likely to perform cash-in. Thus, the value propositions for cash-out agents and cash-in agents need to be considered independently, because theyre often completely different agents. In general, we find that MNOs allocate 25%-50% of the total commission pool for an end-to-end p2p transfer to the cash-in agent.
Theres one other important kind of commission imbalance that can incentivise suboptimal agent behaviour. My colleague Neil Davidson recently noted on this blog that it is possible to pay agents too much for registering new customers. If the registration commission that agents earn dwarfs the commissions that they can earn on an ongoing basis from providing cash-in and cash-out services, agents will sign up customers indiscriminately, without troubling themselves about whether those customers are likely to become users of the service. Some operators have tried to solve this problem by making part of the registration commission contingent on future transactions by the customer.
For more about agent incentives, see our Handbook on Agent Networks: How to Incentivise a Network of Mobile Money Agents.