Financial service providers (FSPs) recognise the value of data to solve their business challenges. In a recent survey that we conducted with 333 providers across six countries in sub-Saharan Africa (Ghana, Kenya, Mozambique, Rwanda, Tanzania and Uganda), 93% indicated they have a good understanding of the business challenges they are trying to solve with data, 62% already have an agreed data strategy, and 45% have a specific budget allocated to data.
The question lies in how and where FSPs can implement data-driven solutions to derive this value and viably serve a wider customer base.
Alternative sources of data are increasingly being used to provide insights on underserved consumer segments. The applications driven by alternative data may, however, be too complicated for FSPs that are relatively new to data-driven decision-making. Fortunately, those who are looking to use more traditional sources of data do have at their disposal several approaches that they can adopt.
As a starting point, traditional FSPs can use internal transactional data to derive new insights to solve existing business challenges. These include challenges such as identifying and segmenting customers or managing agent networks, which can be difficult to resolve in data-poor environments.
Through technical assistance provided to a traditional bank and a mobile network operator (MNO), insight2impact has been able to demonstrate that some of these challenges can be addressed by relatively straight-forward data applications. These applications aren’t entirely dependent on data scientists with advanced skills or an upfront investment in state-of-the-art technology (two of the biggest challenges that our survey respondents reported that they are facing).
If you represent a traditional FSP, it’s likely that the customer data you already have (even if it’s incomplete and imperfect) can be mined to deliver additional value. For customers who undertake informal business activities and transact in their personal capacity, analysing the financial transaction data can provide you with an understanding of the economic sector in which they operate. In certain cases, this could indicate their likely income levels and serve as a proxy for their geographic location.
For example, the economic sector in which a customer operates can in certain cases be determined by the types of account holders with whom the individual transacts. If someone repeatedly transacts with agri-businesses, you might assume they are a farmer. Someone who repeatedly makes payments to a grocery wholesaler may be operating an informal retail outlet or stall. Furthermore, a proxy for the customer’s income can be derived from the volumes and values of the customer’s transactions. The location of a customer can be determined by examining the locations of the agents or branches through which the customer transacts regularly. These indicators can be used to strengthen a sparsely populated customer relationship management (CRM) system and can provide you with additional insights to assist in addressing the challenge of segmenting customers with limited information. This approach requires you to have transactional data but doesn’t necessarily require advanced data analytics skills.
What about distribution optimisation? Mobile money operators sometimes struggle to optimise the placement of agents in rural areas; however, this can be the key to unlocking access to a largely untapped potential customer base. One thing that makes it difficult to optimise placement of agents in rural areas is the lack of proxies for the likely demand of services there. In urban areas, agent network management is comparatively easy, as economic activity is concentrated and there are numerous proxies for the likely demand of services (such as bus stations, permanent markets and registered commercial entities).
However, in rural areas the lack of such aggregation points and lack of detailed information often mean you or your super-agents need to travel around to find potential locations, which can be time-consuming and costly. A viable alternative is to leverage call detail records (CDR) data to map where your rural customers aggregate throughout the day. If you are a mobile network operator, you likely have CDR internally. If you aren’t, you can look at your partner organisations. Analysing these records can assist in optimising the placement of rural agents and can eliminate the need for trial and error as well as intensive market research prior to expanding your network into unfamiliar regions. This approach requires you to have access to CDR data and will need some level of data expertise to execute.
These are just two examples of relatively simple data-driven solutions that can assist you in overcoming the challenges related to viably serving a wider customer base. Detailed instructions on how to apply the CDR data and agent network management solution can be found here. More information on applying network analysis to transactional data can be viewed here.
In producing the case study on customer segmentation, we partnered with KCB Bank in Kenya. The case study focuses on how to identify small-scale farmers based on their transactional data. To produce the case study on rural mobile money agent network management, we partnered with Airtel in Uganda. We are in the process of expanding this portfolio of use cases on data-driven solutions that address financial-services-related business challenges, with an interest in solutions that reach unserved and underserved markets.
To learn more about a solution, replicating a solution or partnering to create a new solution – please reach out to me at firstname.lastname@example.org.