Mobile Money: Even Data Analytics Has Limitations
A lot of attention has been given recently to solving the problem of high inactive account rates among branchless banking customers. Big registration numbers are no longer enough to satisfy the donors, who want to see the impact of their funding, nor enough for the providers, who want to see sustainable revenue generated from new customers.
Photo credit: Francis Minien
Last year, CGAP colleagues released research analyzing transaction-level data of four branchless banking providers to understand how issues like customer demographics, usage in the first month after sign-up and quality of agents impact ongoing customer activity. This research was released on our blog in a three-part series. The research highlighted the important role that data mining and analysis have within an organization to help service providers understand their customers.
Further to this research, CGAP commissioned Sofrecom to analyze the transactional data of the active customer base of a mobile money deployment in a country in West Africa. The objective of the analysis was to create a segmentation of the active user base to inform the distribution, communications and marketing, and product development strategies of the provider.
This deck (in English and French) provides high-level insights coming out of this study. Quantitative data analysis was used to identify the segments among the active user base. With this segmentation, a deep-dive qualitative analysis was undertaken to unpack and validate the findings.
The three main segments that emerged from the study were:
1. Youth: Students less than 20 years old who use mobile money to receive funds from family to pay for school related expenses and use the mobile wallet as a means of limiting expenses.
2. Small business owners: Independently employed people who use mobile money to receive funds from family supporting their business and from customers paying for services. They also use mobile money to send funds to family they are supporting and to pay bills.
3. Sponsors: People that handle high volumes of money in their mobile wallet mainly to send money to family they are supporting. The wallet is rarely used to receive money.
Perhaps the most interesting insight from the study was that it showed that data analytics has its limitations. This particular mobile money deployment has seen a large uptake of “direct deposits.” Colleagues at the GSMA MMU have recently profiled direct deposits and describe it as a customer sending money to someone by depositing the money directly into the recipient’s account instead of depositing money into his own account and subsequently transferring the money to the recipient. By doing this, the sender does not use his mobile wallet to send money to someone else, avoiding the transfer fees. The MMU blog has had an interesting discussion recently about ways providers can avoid this customer behavior.
But what interests us as it relates to this research is that transactions that happen “outside the account” make data analysis somewhat incomplete. Let’s take the case of customer A who is active and sends money to customer B on a regular basis, but does so through a direct deposit directly into customer B’s account instead of through a transfer from his own account. An analysis of the transactions in customer A’s account won’t show any of this activity. Instead of showing cash-in transactions into customer A’s account and subsequent transfers to customer B, the data analysis will only show multiple cash-in and cash-out transactions in customer B’s account. Data analysis will give the mistaken impression that customer A is inactive.
There are ways that close tracking of transactions can reveal this behavior, as the MMU explains. But Sofrecom’s study shows that while quantitative data analysis is a useful first step, it is even more insightful when providers actually go and talk with their customers directly.
----Sarah Rotman, Financial Sector Specialist at CGAP and Claude Robert, Senior Consultant at Sofrecom
Thanks for sharing this, very insightful and interesting. I agree with you that “what happens outside the data” becomes invisible and this is something that we all have to accept as part the limitations to fully understand a phenomenon. But I want to make a distinction between your research interests (when you said “…what interests us…”) and the business problem that any provider will face when trying to address dormancy or inactivity. Let’s look it through the eyes of the provider; on your customer A and B example, customer A is actually inactive because it’s not generating a fee or revenue for the provider. As you said a qualitative approach will help them to get the “why” behind the data and those insights could help providers to tweak their products (or operations) to close those revenue or information loops. I might be biased, but to me the quantitative phase is essential and doesn't conclude when you add a qualitative piece to it. On the contrary, the qualitative insight will give you additional information about the behavioral nuances hidden in your transactional data.
Thanks again, J
This is great. How is the integration of this with market research and statistics in a banking sector