An important part of effectively rolling out a banking agent network - a network of retail or postal outlets that handles transactions on behalf of financial institutions and mobile operators - is the agent location. 

Our project partner, Credibanco VISA in Colombia, is using a georeferencing tool to advise banks as to where large numbers of their target clients are located, and also which retail outlets in that area might make good “corresponsales no bancarios”, as the Colombians call their banking agents.

The process is easy. Based on the bank’s target clientele (e.g., income up to COP 250,000 per month [US$122] and “estrato 2″ reflecting the Colombian economic classification of 0-6, where 0 is poorest and 6 highest income) and preferred location (e.g. high population density, no financial infrastructure, etc.), VISA uses census data, financial infrastructure coverage, and retail information from yellow pages to develop heatmaps which combine the following:

• poverty and income levels 
• population density
• postcode boundaries
• existing financial infrastructure (e.g. branches, ATMs, etc.) and card holders
• stores and other commercial activity
• areas that generate a lot of foot traffic (e.g. bus stations, markets, hospitals, etc.)

Based on the resulting maps, the bank can see in which areas their agent would be most effective. Factors like poverty and income level, population density, and existing financial infrastructure will impact the agent’s future transaction volume; placing agents near bus stops and market areas will make them more visible and increase the likelihood that clients will repeatedly use the agent to conduct transactions.

On the picture you see the final heatmap. Red areas show neighborhoods with great opportunity to reach the bank’s target clients; the mountains refer to population density. If you would like more detail, please send me an email and I can forward you VISA’s complete analysis of Ciudad Bolivar, a poor neighborhood of Bogota, Colombia.