Before launching a mobile money service, it behoves operators to think carefully about how to prevent abuse that would harm customers. Such foresight helps assure regulators that the service will be safe and in turn safeguards the operator from reputational risk that would arise from its failure or widespread fraudulent use.

But customers and agents aren’t the only ones who need to be protected. Increasingly, operators have begun to recognize that they too require protection from abuse – otherwise, they face the prospect of watching their revenues erode as customers avoid fees and their costs rise as agents maximize commissions.

The GSMA recently convened a group of operators to share their experiences with abuse. From our session, it became evident that operators at highest risk are those offering domestic money transfer services using tiered pricing or commission models. Still, it was clear that every operator – whether they’re offering bill payments or p2p transfers using a simple tariff model – needs to think carefully about how a customer or agent could game the system and affect their business model. Below we describe three of the most common ways in which customers and/or agents have attempted to abuse mobile money services.

Splitting Transactions

A cash-in/cash-out or bill payment transaction has been ‘split’ when it’s divided into two or more smaller ones. This can be triggered by a customer attempting to minimize their fees (i.e. “If I withdraw my $50 in two $25 tranches, I only pay $1 instead of $2 in cash-out fees”), or by an agent seeking to maximize their commissions (i.e. “If I tell this customer to make two simultaneous withdrawals, I’ll earn $2 instead of $1 in commissions”). Generally speaking, this behaviour is prompted by a pricing or commission structure that varies based on the size of the transaction.

Whether a transaction is split on the instruction of a customer or an agent, the outcome is equally troublesome: the operator’s margin shrinks. The simplest solution to this form of abuse would be to abolish pricing or commission tiers altogether, but this option has largely been ignored since neither alternative to such tiers is particularly appealing: a percentage-based commission is confusing and can result in agents being under-paid for low value transactions and over-paid for high value ones, while a flat commission per transaction has the reverse effect. Instead, mobile operators have carefully scrutinized their pricing and commission tiers to identify opportunities for customers or agents to game the system and put in place processes and/or technology to identify and deter it. For instance, one African operator generates a daily report identifying customers who have made two withdrawals from the same agent within 5 minutes, a clear indication of splitting. And in Asia, an operator has implemented a rule that precludes its agents from making multiple payments toward one account number in one day.

‘Direct Deposits’

A ‘direct deposit’ occurs when the customer initiating a P2P transfer hands an agent cash, but provides them with the mobile number of the recipient rather than their own in an effort to avoid paying a transfer fee. While agents are ultimately the enablers of this form of abuse, they may be complicit (i.e. “hey, I can help you avoid the $0.50 transfer fee – just pay me half of the P2P transfer fee you would have paid in cash”) or equally they may not be aware it’s even taking place (i.e. the customer might say “the mobile number you should credit is…”).

Operators have quickly recognized the importance of protecting themselves against this form of abuse and implemented a number of prevention and detection measures. First, they have attempted to educate customers: on tariff posters, operators often note that customers are only allowed to deposit funds into their own account. Second, they have again relied on technology. In this case, operators look for instances of a customer depositing and then withdrawing funds from the same MSISDN number within 5 minutes – this has proven to be a reliable proxy for direct deposit abuse. And finally, operators have adjusted their pricing models to minimize their exposure to this form of abuse. For instance, one African mobile operator recently changed their tariffs to become less reliant on transfer fees as a source of revenue and instead to earn the bulk of their revenue from cash-out fees.

Unfulfilled Transactions

An unfulfilled transaction occurs when the customer provides an agent with cash, whether it’s to pay a bill or load their e-wallet, but the agent does not credit that value to the customer’s account. The risk of this occurring goes up when a busy agent collects cash from customers throughout the day and then, when their store is less busy, performs all of their transactions in bulk. While this might seem to an enterprising agent to be an optimal way to schedule his workflow, it exposes customers to the risk that an agent might make a mistake, or, worse, deliberately fail to record the transaction, pocketing the value of the cash-in. This can ultimately lead to lost revenue and unsatisfied customers for the operator.

Education has proven to be the most effective way of preventing instances of unfulfilled transactions. Prudent operators have helped agents appreciate the importance of performing transactions in real time, and equally ensured that customers understand the importance of waiting until they have received a confirmation SMS before they assume that the transaction has been completed.

In this case study, we examine the key elements of Zain’s strategy for deploying Zap in East Africa, including their approach to organizational design, service design, marketing, distribution, and bank partnerships. We also provide a snapshot of the initiative’s current status, including their successes (i.e. registering 4 million users in Tanzania) and challenges (i.e. underinvestment in certain key areas).

This will be an important and interesting year for Zap. Be sure to read this case study to understand how the deployment is positioned, and as always we look forward to your feedback.

Something particularly interesting about the deployment is their decision to launch using a shared platform. To understand more about their approach, I recently had a conversation with Qasif Shahid and Mazahir Ali Sayeed from MCB, and Hannes van Rensburg from Fundamo.

GSMA: What is the Shared Platform?

Hannes: it’s a mobile banking platform hosted by Access Group, our partner in Pakistan. They’ve deployed the platform, which is provided as a managed service to customers like MCB. Currently, there are banks as well as MNOs sharing this platform, and each party has launched a service using their own model, wisdom and insight.

This isn’t the only shared platform in the world – Fundamo has deployed them in other markets as well. For instance, groups may choose to deploy a platform in one country and then allow OPCOs in other countries to run off it.

But as far as ‘in-country’ shared platforms go, one of the reasons Pakistan has been so successful is that the shared platform serves 60-75% of the mobile banking deployments – so there’s a critical mass of participants on board. If they work together to create interoperability, then you create momentum and incentives for others to join. A shared platform that serves, say, 30% of the market is less likely to be successful from this point of view.

GSMA: What are the benefits of using a shared platform?

Qasif: For MCB, its basically been a way of fast tracking the entire speed of execution because we haven’t had to worry about the platform, its scalability, or connectivity of the platform with other mobile operators and the other things that go along with this area of technology. Instead, we’ve been able to introduce a completely new service in just 6 months. And not just introduce the service – we’ve also grown quickly because we’ve been able to focus on educating customers about the service, creating excitement and so on.

So speed to market and ability to focus on a sound customer proposition have been the benefits from a commercial bank like MCB’s perspective. But as we go down the opportunity pyramid, you see microfinance banks and other financial institutions struggling with a lot of things. And what the shared platform does is allow multiple banks to use one platform to make the business case make a lot more sense for these types of organizations.

We also feel that interoperability is an important element. It all sounds very good that everyone is going to make their own platform and that they’ll connect the platforms in such a way that it will work – but imagine a world where you are all sitting on the same platform, including banks, mobile operators and microfinance banks, and thus interoperability is simply a by-product that comes out of this approach.

Hannes: For us, a bank or MNO’s decision to select a shared platform is to some degree economic, but it’s more a product of go-to-market strategy than anything. It provides for quick speed to market, flexibility around service design, and ensures resources will be available to focus on educating the customer. Of course for this to be possible there needs to be a willingness to participate with others on the platform.

GSMA: How central is interoperability to the essence of a shared platform?

Qasif: Interoperability is at the heart of all this. The consequence of you being on the shared platform is that you’re interoperable. There are 5 MNOs in Pakistan: imagine that you could only make calls to one of them – its just not good enough. So if I have multiple banks and multiple MNOs all residing on the same platform, then the potential take-up of mobile money will enter a new dimension completely. If you look at the internet and mobile telephones, they all had limited transformational impact on the world at first because they were not interoperable, but when this changed they all went big.

GSMA: What have the early experiences been for MCB?

Qasif: We launched the service in June 2009 after a 3 month pre-launch period with our 15,000 employees in which we mandated that they all use the service and provide us with feedback.

MCB has 6 million accounts out of the 30 million in Pakistan, so we’d have a lot to lose by missing out on mobile banking. Today we are registering about 250 to 300 customers per day for the service – these are our existing customers who we’re registering to use the mobile channel. This is a healthy rate of growth. So far we’ve done over 500,000 transactions and feel that a lot of our success to date is because we’ve not been concerned with management of the infrastructure, scalability and connectivity that comes with it. Instead, we’ve been able to start quickly and worry about the ‘packaging and functionality’. It’s our belief that the platform and the network are not sufficient differentiating factors – the real differentiating factor is the organization’s ability to construct processes to which you can convert customers, educate them, and transform their behaviour.

GSMA: So how have you done this at MCB? Particularly, how have you driven adoption?

Qasif: Today, if you wanted to sign up for the service, you can enter through several channels. When you use an ATM, swipe a card at a POS, call the call centre, walk into a branch, or visit the website, each of these channels will contribute to driving you to adopt the service. Organizationally, MCB has put innovation at the middle of all of our channels. Banks sometimes make the mistake of putting innovation as its own team within an organization – it really needs to span all channels.

Hannes: I’d add that MCB is so successful because they’re clear about their channel strategy – why are they doing this? What are the metrics? Mobile banking is not a single strategy – it fits into a bigger strategy and where other banks are concerned with their platform, MCB is doing well because they have a coherent strategy of which mobile banking is one part.

GSMA: Thanks, Qasif and Hannes.

Before addressing why cash is so painful and how Zap is helping, let’s quickly review the distribution model Coke uses to get their product onto the shelves of retailers. In Tanzania, there are a handful of distributors who regularly supply about 400 ‘mini distribution centres’ (MDC) that are scattered across the country. These mini distribution centres are then responsible for supplying Coca Cola product to between 80 and 300 retailers each. Coke’s model is simple and clearly effective, but their reliance on cash for settlement at various phases – whether it’s the retailer paying the MDC truck driver in cash, or the MDC branch paying the Coke distributor’s truck driver in cash – creates serious challenges.

Consider the scene at an MDC when their truck returns from making deliveries to retailers. The inventory has all been delivered, but now the driver is carrying cash – and lots of it. This makes them an instant target for thieves upon their return to the MDC, which offers little in terms of security or privacy. A similar challenge occurs – but on a much bigger scale – when the Coke distributor truck comes to re-stock an MDC. As the distributor’s truck approaches, the MDC owner patiently waits with all of the cash he’s accumulated since the last delivery – and as the crates are unloaded, the driver and owner sit together and slowly count hundreds – or thousands – of notes.

Clearly, the use of cash creates security risks in Coke’s distribution model. But it also creates inefficiency and tension since drivers are often paid per delivery run completed and are not eager to sit around counting notes, first at the MDC and then again at the distribution headquarters.

To address these challenges, Zain and Coke are conducting a pilot with about a dozen MDCs who have begun paying their distributor via Zap. Thus, when an MDC delivery truck receives cash from a retailer, they can either deposit it into a Zap account via an agent at any stage of their delivery run, or the owner can take the entire sum to a Zap bank partner at a later stage. Then, when their distributor comes to re-stock them, the settlement is done safely through Zap in a matter of seconds. This eliminates the security risk of carrying cash, as well as the need to spend time counting notes at the MDC and the distributor’s headquarters during a delivery.

Today we’d consider this a B2B application of mobile money, but that’s not to say it won’t have implications for the “C’s” at some point in the future. An ideal, though tricky, next step for this initiative would be to encourage the retailers who currently pay Coke in cash to transition to Zap, and for them to equally begin accepting Zap for retail payments from customers. Of course, making this transition is not easy and would require other building blocks to be in place, but Zain’s vision does offer a glimpse into how one cashless ecosystem might develop and function.

Zain isn’t the first mobile operator to conceive of providing this type of service to Coke-like customers. But where others have stumbled, they’ve had some early success for some small but important reasons. First, Zap is integrating with multiple banks. This is important to customers like Coke because the MDCs they work with often have lines of credit and are obligated to route money through a particular bank under the terms of a loan. Thus, with multiple bank partners, the number of MDCs eligible to use Zap increases. Additionally, Zap has a stated goal of providing ‘much more than money transfer’ – and view corporate customers as a key part of their ecosystem. Thus, they are willing to spend the time and money needed to customize the solution for these business partners, particularly ones as strategically valuable as Coke.

We’ll be watching this partnership closely to see how it grows in 2010.

The active user rate plays an important role in a mobile operator’s business model – if it’s unexpectedly low, it impacts cost of acquisition as well as the value a deployment can provide to its agents. Recently, we had conversations with five different deployments who shared their active user rates with the GSMA. To protect confidentiality, I won’t reveal which deployments – or even which markets – the numbers come from. Still, the figures should spark some discussion (and hopefully creative solutions) from the community.

Of the five deployments, one had an active rate of 60%, and the other four were all below 30%, with the lowest of the bunch coming in at 10%. Clearly, there’s room for improvement across the board. But to begin thinking about how we can improve the active user rates in any market, we first need to understand why they’re low in the first place. There’s an endless list of possible reasons – and please add your ideas below – but I’d like to start the conversation by presenting two that I think are particularly prevalent.

“Some of our agents do registration, and others do cash in/out”

A number of deployments around the world have created agent networks with two categories of agents: one to register customers, and another to facilitate cash in/out. In some cases, this structure has contributed to low active user rates for a few reasons. First, by creating a category of agents whose sole responsibility (and source of income) is to register new customers, it can be difficult to prevent them from selling a bit too eagerly to prospects with no actual demand in an effort to earn a commission. In this case, it’s less about a customer registering, and more about a customer being registered. Second, customers who register at dedicated ‘registration agents’ face an immediate barrier to use: finding an agent. Today in Nairobi this is easy – but in most early stage deployments, it can be a difficult task. And third, because registration agents typically earn most of their commission from the act of registering a customer, it’s difficult to persuade them to invest time in education.

“It’s our strategy: Phase One is acquisition, Phase Two is activation”

Another reason for low active rates is the philosophy that mobile money should be rolled out in two distinct phases. Disciples of this philosophy use phase one to register as many customers as possible and then later, in phase two, think about finding ways to get them transacting. If the active user rate equation has a denominator (number of registered users) and a numerator (number of active users), then this philosophy calls for a focus on the denominator in phase-one, and then a shift to focus on the numerator in phase-two.

We’re seeing this strategy play out in many African markets where operators are in the midst of implementing prepaid SIM registration. Operators are leveraging the unique touch-point with customers that SIM registration provides as an opportunity to promote mobile money. Of course, just because an operator activates an e-wallet for a customer who has gone through KYC, it does not necessarily mean that customer will ultimately be an active user. But, critically, it does give the operator an improved likelihood of success by eliminating the registration barrier and enabling the delivery of promotions. Still, the ultimate success of this strategy will hinge on phase two: giving customers a compelling reason to start transacting – and making sure that the ecosystem needed to facilitate transactions is still intact.

But why does the active user rate matter so much in the first place? People are always curious to hear the figure, but how exactly will a low active rate impact the rollout of a deployment? Well, it depends on who within a mobile operator you ask.

If you’re asking the head of distribution, he might say something like this: “I don’t care how many customers you register because only a fraction of them end up giving me what I need – transactions that will generate income for my agents. If customers don’t start transacting, my agents will lose interest and we’ll have no distribution network.” It’s clear then that the head of distribution cares more about active users than registered users.

If you’re asking the head of finance, she might say “in your business case you estimated the lifetime value of a customer to be X, which justified a cost of acquisition of Y. However, with so few customers transacting, the average lifetime value is too low to justify our COA.”

The trick, then, is to first have realistic estimates not just of registered users but of the number of active users, and to constantly be thinking of ways to drive usage. And simple things can make a big difference. An Asian deployment facing a low active rate partially due to having multiple categories of agents found that simple follow-up calls to educate customers got them transacting. An African deployment facing the same challenge found that encouraging their registration agents to perform activations in front of cash in/out agents helped to improve the active rate by clarifying to customers how they can start transacting. And in many deployments, operators have already begun stipulating that registration commissions will be paid in multiple tranches – some amount for registration, and the rest held for when a customer completes a certain number of transactions.

Safaricom and other veterans can look back at their historical figures and say that none of this should be surprising. The active user rate is something that requires attention, but can be addressed as a deployment matures. To help advance the thinking, please feel free to share your ideas below on ways that deployments at various stages of maturity can improve their active rates.

Q: I’m not from Latin America – will MMS be relevant to me?

A: Yes. While this event is being held in Brazil, it will feature global content. Whether you’re from Africa, Asia, North America or the North Pole, there will be relevant content for you at MMS.

Q: What if I don’t focus on the unbanked segment? Is the event still relevant?

A: Yes. There will be content on the programme that’s relevant to people interested in both banked (i.e. general programme) and unbanked markets (i.e. MMU working group).

Q: I go to many events. Will I be hearing the same things from the same people in Rio?

A: No. MMS will feature fresh, new speakers. Further, any familiar names who have spoken recently will be required to present new and interesting content.

Q: I want to share content / make a presentation at Mobile Money Summit. What should I do?

A: You’re in luck – but you need to act fast. The call for papers closes on March 5th. Click here for more information.

Q: I’m a technology vendor (or other service provider) and I’d like to find a way to reach my target market. What do you suggest that I do?

A: Click here to learn about advertisement, exhibition, or sponsorship opportunities at MMS.

Q: What was the 2009 Mobile Money Summit like?

A: 2009 was a great year, and we’re confident 2010 will be even better. Last year the event attracted 457 attendees from 68 countries. 63% of these attendees were C-Level, VP, Board, or Directors. There were 42 speakers last year and 82% of attendees would ‘strongly’ or ‘definitely’ recommend the event to a colleague. Check out some of the presentations from 2009 at www.mobileworldlive.com.

Q: We are trying to launch mobile money, but are having difficulty working to get approval from our financial services regulator. Will there be any content at MMS to address this?

A: On Monday 24 May we’ll host a free regulatory executive training session. This full day session is open to mobile operators, and is designed to help them better understand the ways they can influence their financial services regulator. To provide direct and valuable feedback, a regulator will take part in the session. The most appropriate person to attend this session would be the mobile money regulatory lead contact from the MNO.

Q: I’m ready to buy a ticket to attend. Where do I go?

A: Stay tuned – in the coming weeks, this will be possible at www.mobilemoneysummit.com.

The roundtable was attended by MTN, Orange, Zain, Camtel, Citibank, Afriland First Bank, BEAC, BCEAO, Central Bank of Kenya (thanks to the support of the Alliance of Financial Inclusion), Bank of Ghana, CGAP and GSMA.

So why did we choose to host our first regional regulatory roundtable in Cameroon? If you look at our deployment tracker, it’s clear that East and West Africa are hotbeds of mobile money activity, but so far no deployments have been launched in Central Africa – this in spite of the fact that MTN, Orange and Zain all have strong footprints in the region. Our hope is that this will change in 2010, and if Tuesday’s session was any indication, the outlook is positive.

I’d like to highlight two promising themes from the day:

Mobile Operators and Banks Working Together

MTN, Orange and Zain all made brief presentations describing their models, and it was clear that each of them have logically divided responsibilities between bank and MNO based on which party has the relevant expertise. It became clear to the participants (and regulators) that there is a clear win-win situation both for banks and MNOs (not to mention the social and economic benefits for Central Africa) when they work together.

Dialogue – Curiosity From All Sides

We left a great deal of time for discussion on the agenda, but even still it wasn’t enough. Regulators were keen to share their perspectives and ask candid questions to their peers from other markets, as well as the banks and operators in the room. It’s clear that there is a huge desire to exchange knowledge, particularly between countries that share similar contexts and challenges.

A lot of the day’s value came from creating dialogue between the key stakeholders on the specific issues that are relevant to Central Africa as a precursor to (hopefully) approving the launch of mobile money. There is an incredible amount of potential for mobile money in Central Africa – but, like any other region, there are questions about the service that must be answered first. Tuesday was an important step in the process.

We’d like to thank our co-hosts, MTN and Citibank for helping to make this roundtable possible, as well as BEAC for actively participating in and contributing to our first regional regulatory event.

The focus of the day was agent distribution, and conversations ranged from individual agent economics to the best practice for structuring an agent network, to the role that scratch cards might play. After some introductory remarks by Seema Desai, the day got started with a presentation on agent economics in Brazil by CGAP’s Mark Pickens. You can find his summary and presentation here, but I found a couple things particularly interesting:

- Agents handle 2.4 billion transactions per year
- Agents in Brazil make just $5.17 in profit per day, or 4 cents per transaction
- Agents do 166 transactions per day, or 1 transaction every 3 minutes

After Mark, I provided some learnings provided to me by Cambodia’s WING. Their model is particularly interesting and they’ve got some hard earned experience tackling key issues that many operators will face early on in their launches. Much of the conversation oriented around WING’s decision to use ‘Pilots’ for registration, and WING Cash Express Agents for cash in/out. Some operators, like MTN Uganda, have used a similar approach (although WING calls their staff ‘field registration agents’ rather than ‘Pilots’, so the learnings were timely. To help put WING’s model in context, we used M-PESA’s agent network as somewhat of a foil. Comparing the agent networks, it was interesting to see that there are two main differences: first, whereas M-PESA has one uniform agent that provides registration,c ash in and cash out, WING has two categories of agents – one category that does registration, and another that does cash in/out. Second, whereas M-PESA’s network of today has multiple tiers, WING’s had just one at launch – a second one was added about a year later and this looks like the tiering path that many models are following (including M-PESA).

Another highilght of the day was a discussion panel that included representation from Zain, Grameenphone, Roshan and Globe. The focus of the panel was on agent networks, and one learning that I found particularly interesting was the approach operators on this panel have taken to structuring agent incentives. Globe, Grameenphone and Zain all provide their agents with some degree of input into the ultimate price of the service, and thus, the agent’s commimssion. This approach varies notably from Safaricom’s fixed tariff structure, which is often touted as contributing to the consistent customer experience that fuelled the growth of the service.

I’ll share more in the coming days but thought I would share my highlights from a day where many people contributed to the learnings shared. Keep a watch out for the presentations from the day.

Case Study Presentation at MWC

Not surprisingly, the day was filled with learnings… and numbers. I’ve taken a moment to share what I thought were the seven most interesting ones:

1: New deployment launched in India. Today, Nokia Money launched in India and this will certainly be a deployment to watch.

2: The number of years that it takes for a mobile money ecosystem to make money and become sustainable, according to Cenk Sedar, Director, Vodafone.

3: The number of developing markets that Telefonica will have launched in by the end of 2010. This follows on the release last week by Trivnet of their partnership with the operator.

13: the number of minutes that it takes for a customer to report their phone stolen, according to a study conducted by Telefonica. This compares to 1-2 days for a bank card.

500,000: the number of transactions that Pakistan’s easypaisa has processed in the last four months.

3 million: the number of customers that M-PESA in Tanzania has reached.

8 million: the number of customers that M-PESA in Kenya has reached.

And finally, we couldn’t put a number on this one, but it merits sharing. According to Vodafone’s Cenk Sedar, M-PESA will process more transactions per year than Western Union by the end of 2010. Truly a reflection of their growth.

Tomorrow is the Mobile Money for the unbanked Working Group, and I’m looking forward to the exchange of learnings. More to come throughout the week.

Click here for day two.

Launched in 2005, True Money is now used by 6 million customers. The system processes over USD$900 million in electronic payments and 120 million transactions per year. This doesn’t match the US$3.5 billion in p2p payments M-PESA processes per year, but it does suggest that the model has gained some traction.

What’s most interesting is just how unique True’s approach has been. Whereas M-PESA has scaled remarkably well as a money transfer offering, True Money has gained its traction as a payments offering. That is, their scratch cards, e-wallet and 8,000 ‘True Money Express’ agents form the basis of a system designed to process True Group bill payments, 3rd party bills, and prepaid bills from niche issuers.

True Money and M-PESA differ in many ways, but are both success stories in their own right. Indeed, True’s success to date vividly illustrates how market conditions, customer needs, and operator assets shape the nature of the mobile money opportunity in every market.

Read the case study and share your thoughts below.